flow-workflows 0.76.0 → 0.78.0

package/adapters/codex/CORE.md

@@ -17,7 +17,7 @@ Every `$flow-*` command assumes these rules. They are stated once, here, so a co
 carries what is specific to its phase. Read this once per session; a command that says "load
 `flow-core`" means this file.
 
-**This file belongs to flow `0.76.0`.** Compare it once, at the start of the session, against
+**This file belongs to flow `0.78.0`.** Compare it once, at the start of the session, against
 `version` in `~/.claude/flow/.claude-plugin/plugin.json`. The two differing means the session
 is running a **mixture** — the commands from one copy of the plugin, these shared rules from another
 — which is exactly what happens when a branch or a release candidate is loaded over an installed
@@ -98,8 +98,7 @@ a command relies on is simply absent in that session, and nothing else will repo
      "general-purpose", "phase": "review" }`. Asking here would break the never-a-question contract
      (§2); recording keeps the decision readable afterwards, and `$flow-doctor` aggregates it.
 
-  **The list, and it is short on purpose**: `agents.<role>` when a panel is about to improvise that
-  role on an M/L or sensitive diff · `quality.review_depth` after a review's cost line went over
+  **The list, and it is short on purpose**: `quality.review_depth` after a review's cost line went over
   `agents.budget_max` · `data.volumes` or `data.explain_cmd` when the query duel came back
   schema-only · `quality.functional_check` when `validate` found a runnable app and no way to drive
   it · `tracker.start_cmd`/`done_cmd`/`abandon_cmd` the first time a ticket could have been
@@ -157,8 +156,8 @@ improvises, `models.workers` for the parallel fan-out rounds, and `models.superv
 bounded waits of §6.5 — the last two falling back to `models.agents`. Empty, absent, or no section
 → everything runs on the model you were launched with, and you say nothing.
 
-- Pass the resolved value to every subagent you launch, **except** an agent named in `agents.<role>`,
-  which keeps the model its own definition sets.
+- Pass the resolved value to every subagent you launch, **except** an agent named in `agents.<role>`
+  or picked from the installed ones (§6.6), which keeps the model its own definition sets.
 - **You cannot switch your own model**, so no key names the phases the main agent performs itself.
   A phase that wants another model says it in one line at the handoff (`this step reads better on
   <value>, you are on <current>` → `/model <value>`), records it in the artifact, and continues.
@@ -296,6 +295,9 @@ prose describing a file is otherwise obeyed to the letter.
     "",
     {"ref": "#1", "text": "batch read sources", "mark": "wait", "link": "https://gitlab.com/…/merge_requests/9977"},
     {"ref": "#2", "text": "per-event and per-recipient counters", "mark": "current"},
+    {"ref": "1–2", "text": "counter columns + migration · repository query", "mark": "done", "indent": 4},
+    {"ref": "3", "text": "per-recipient aggregation in the use case", "mark": "current", "indent": 4},
+    {"ref": "4", "text": "expose both counters on the endpoint", "mark": "pending", "indent": 4},
     {"ref": "#3–#6", "text": "channel map · use case · document detail · route", "mark": "pending"},
     "",
     {"ref": "Now", "text": "unit suite and the test agent over #2", "mark": "info"},
@@ -307,10 +309,10 @@ prose describing a file is otherwise obeyed to the letter.
 }
 ```
 
-- **`mark` says what a line is**; the reader draws it. `done` · `current` (at most one) · `pending` · `wait` (waiting on someone else: an open MR/PR, a user decision) · `block` · `info`. Marked lines form an aligned column: symbol, `ref`, text, link pinned right. Do not set `style` on a marked line — except `mark: "info"` + `style: ok|warn|error` when the colour *is* the information (a monitoring verdict).
+- **`mark` says what a line is**; the reader draws it. `done` · `current` (at most one per block) · `pending` · `wait` (waiting on someone else: an open MR/PR, a user decision) · `block` · `info`. Marked lines form an aligned column: symbol, `ref`, text, link pinned right. Do not set `style` on a marked line — except `mark: "info"` + `style: ok|warn|error` when the colour *is* the information (a monitoring verdict).
 - **`ref` need not be a number**: `#1`, `#3–#6`, `Now`, `Next`, `Decision`. Column width is per block; blocks are separated by blank lines — use them so the MR/PR train and the labels below do not drag each other wide.
 - **`link` is a field, never a URL inside `text`** — and **every line that names an MR/PR carries it**: the train entries, a blocker about somebody else's MR/PR, a `Now` or `Next` about the open one. The URL is in `meta.json.mrs[].url` or in whatever recorded that blocker; a bare `!10327` typed into `text` is a number the reader cannot make clickable, which is the whole point of the field. No URL recorded anywhere → the number in `text` and nothing else, and go and record it.
-- **Order:** (1) work title (`style: title`, no mark); (2) the MR/PR train, one entry per `meta.json.mrs[]` with `ref` `#n`, short title, real-state `mark`, `link` when there is a URL — not-started entries collapse into one `#a–#z` `pending` line; omit the block without `mrs`; (3) `Now` — what is running, the one fact `meta.json` cannot hold; (4) `Next`; (5) `Decision`, `mark: wait`, **only** when parked on the user — over the tools it travels with `attention: "wait"` (§4.2), and both are cleared together; (6) blockers, `mark: block` (a sibling repo with `contract_handoff: pending`, a red pipeline, an unmerged dependency).
+- **Order:** (1) work title (`style: title`, no mark); (2) the MR/PR train, one entry per `meta.json.mrs[]` with `ref` `#n`, short title, real-state `mark`, `link` when there is a URL — not-started entries collapse into one `#a–#z` `pending` line; omit the block without `mrs`; (2b) **the plan steps of the MR/PR being built**, right under its train entry — only while `build` or `fix` is working through a written plan (`$flow-feat-build §2.0ter`, `$flow-bug-fix §2.0bis`); one line per step, `ref` its number (`1`, `2`…), `indent: 4`, `key` `step-<n>`, `mark` `done` · `current` · `pending` from the plan's ticks in the phase artifact — never ahead of them; past three finished steps they collapse into one `done` line (`1–4`, `key` `steps-done`), and past three pending ones the tail collapses the same way (`key` `steps-pending`), so the block stays at five lines or fewer; drop the block when the MR/PR's build ends. This is the step list the user watches: a harness's own task list, where there is one, is a bonus on top — the panel is what every harness can feed; (3) `Now` — what is running, the one fact `meta.json` cannot hold; (4) `Next`; (5) `Decision`, `mark: wait`, **only** when parked on the user — over the tools it travels with `attention: "wait"` (§4.2), and both are cleared together; (6) blockers, `mark: block` (a sibling repo with `contract_handoff: pending`, a red pipeline, an unmerged dependency).
 - **Rules:** `phase` is the phase you are **running now** (not `meta.json.phase`, which advances only at Close). `header: true` means ticket, type, phase and age are drawn by the reader — do not repeat them in `lines`. Keep it under ~14 lines. Every fact from `meta.json` and the artifacts — an invented MR/PR state is worse than a blank panel. Write in the language of the work's artifacts.
 
 ### 4.2 Over the terminal's MCP — `panel_set` once, `panel_patch` from then on
@@ -385,6 +387,10 @@ that never asked, because they answered it and the screen says it did not land;
 nobody clears keeps that tab claiming them for ever. The same applies to a blocker that lifted and
 to a `wait` line whose MR/PR merged.
 
+**(f) Each plan step that lands** (the (2b) block): one patch, the moment its tick goes into the
+phase artifact — that step to `done`, the next to `current`, `Now` moved on. Not batched at the end
+of the MR/PR: a list that jumps from nothing to finished told the user nothing while it mattered.
+
 ### 4.5 `meta.json.cost[]` — what a phase measured, when it could
 
 A phase that can measure what it spent appends one entry: `{ "phase": "review", "at": "<ISO>",
@@ -549,6 +555,61 @@ moving.
 than to run — the brief, the round trip and the report are the whole of the saving. And anything you
 must *react to* mid-wait is not a wait at all: it is the work, and it stays here.
 
+### 6.6 — Which agent a role resolves to
+
+**The harness already tells you which agents are installed**, each with a line on what it is for,
+and choosing the one that fits a piece of work is something you do well without being told. A rule
+that answers every empty role with `general-purpose` throws that away: a repo with a frontend
+specialist installed and no `agents.frontend` got a generalist told to act like one. So wherever a
+command says *"`agents.<role>`, empty → `Agent general-purpose` with the role"*, it means this:
+
+1. **`agents.<role>` names an agent** → that agent, as always.
+2. **Empty** → the installed agent whose description fits the role **and this work's stack** best —
+   read the list your harness gives you, never a guess at a name. None fits clearly →
+   `general-purpose` with the role in the prompt, as before. An agent that fits the role but not the
+   stack (a PHP reviewer on a React diff) does not fit.
+
+**`agents.selection`** decides how far your judgement reaches:
+
+- **`configured`** (empty → this) — the two steps above. A role you named is never second-guessed.
+- **`open`** — a named role is a default, not an order: for a piece clearly outside what it
+  covers, or where an installed agent fits this piece plainly better, take that one. And a piece
+  **no role covers** — a UI component in a work whose roles are all backend — may go to the
+  specialist that does cover it. Every substitution is **one line in the phase artifact**: role,
+  the agent configured, the agent used, and why. Unrecorded, a panel that ran with other agents
+  reads exactly like the one you configured.
+
+  **Nothing installed fits → create the specialist for this work.** An installed agent is, in the
+  end, a brief with a name; what made it good is the brief. So instead of *"act as frontend"*, write
+  the one an expert would carry — the stack and its versions as this repo uses them, what to check
+  in this piece, the mistakes typical of it, and the report contract of this section — and launch
+  `general-purpose` with it. Nothing is written to the harness's agent folders: the specialist lives
+  for the brief, and it works on every harness, because every one has a general-purpose agent.
+  Record it in `meta.json.specialists[]`:
+
+  `{ "name": "react-form-reviewer", "role": "frontend", "phase": "review", "brief": "<the full text
+  you sent>", "useful": null, "status": null }`
+
+  Once its result is in, set `useful`: `true` when something it returned was kept — a finding
+  accepted, a piece merged into the diff — `false` otherwise. Only a useful one is ever offered for
+  saving, and only **when the work ends** (`$flow-feat-ship` §5.1), never mid-phase: by then the
+  MR/PR is open and it is plain which of them earned a place among your agents.
+
+Three things this never touches, in either mode:
+
+- **The review roster** (`quality.review_skill` / `quality.reviewers`) runs as defined, with no
+  substitutions: its value is being the same panel every time, so a result can be compared.
+- **The agents that are `general-purpose` by design** — challengers, skeptics, the completeness
+  critic, the blinded contract check. They are generalists *on purpose*: a specialist brings the
+  assumptions the blinding is there to strip. A step that writes `general-purpose` with no role
+  behind it means exactly that.
+- **The budget.** A picked agent counts against `agents.budget_max` like any other. Like a named
+  agent it keeps its own model (§1), and `models.agents` reaches it only when it has none.
+
+Record what you picked where a default would be recorded (§0): *"`agents.frontend` empty → used
+`react-expert` (installed)"*. That line is also the evidence for writing the role down when the
+next stop offers it.
+
 ## 7. Deferred work — `followups[]`
 
 Every chain in this plugin reaches moments where the right call is **not to do it here**: a build

package/adapters/codex/skills/flow-bug-fix/SKILL.md

@@ -95,7 +95,7 @@ Save it in `04-fix.md` under `## Plan`, right after the brief. **A fix that was
 
 - Apply the minimal fix targeting the finding of `03-investigation.md` (or the diagnosis if investigate was skipped).
 - Sensitive area (authentication, payments, sensitive data) → consult the `agents.architecture` agent from FLOW.md to confirm the correct layer; empty → check against `conventions` in FLOW.md.
-- More than 2 fix steps → the §2.0bis plan is already written: track its numbered steps with `TaskCreate` where the harness offers one — looked for as `$flow-feat-build §2.1` says, since it may be deferred or named otherwise — otherwise tick them off in `04-fix.md` where they stand and say so once in the first step report. One task per step, same order, marked `in_progress` and `completed` as they go — not batched.
+- More than 2 fix steps → the §2.0bis plan is already written: tick its numbered steps off in `04-fix.md` where they stand, and show them in the live panel's step block (flow-core §4.1 (2b), patched per §4.4 (f)); a harness task list, if there is one, goes on top as `$flow-feat-build §2.1` says. One task per step, same order, marked `in_progress` and `completed` as they go — not batched.
 - **Comment discipline**: comment only a *why* the code cannot say (a non-obvious constraint, the reason for a workaround, a subtle invariant); match the surrounding file's comment density. **Never write the ticket ID or "fix for #N" into a code comment** — that lives in the commit/branch/MR-PR.
 - Keep the log updated while editing.
 

package/adapters/codex/skills/flow-bug-ship/SKILL.md

@@ -186,6 +186,7 @@ nothing was measured. A slowness bug is the case where this comment *is* the pro
 - **Cross-repo reminder**: for each `meta.json.related_repos` entry not `done`: *"you've shipped the `<this-repo>` part; `<repo>` still needs: `<scope>` → go there and run `$flow-bug-start <TICKET>` (or `$flow-feat-start`)"*. flow only reminds; this is not a hard gate.
 - **Cross-repo contract handoff**: for each such entry with `contract_handoff: "pending"` — the fix changed a surface that sibling consumes — hand the new shape over as `$flow-feat-ship` §6.3 (publish the **literal** shape to the tracker ticket, mandatory preview in every autonomy mode, then record `published → <location>`; versioned-file fallback when there is no tracker). A bug flow has no §"External contracts": take the literal from the fix itself — the actual key, code, route or event the code now emits — read it off the code, never from memory.
 - **Follow-up triage — `$flow-feat-ship` §6.4, in full and unchanged.** Read `meta.json.followups[]`; nothing `proposed` → skip silently. Otherwise §6.4 step 0 first — **3 or more candidates go to a skeptic** whose job is to refute that each deserves a ticket, and only the strongest `tracker.followup_ask_max` (empty → **2**) reach the user (flow-core §7) — then one batched `AskUserQuestion` (**Do it** / **Not worth it** / **Later**; `decision` entries asked as their question, `tooling` and everything `log`ged never asked), the accepted ones get their tracker issue created (outward-facing → **asks in every mode**), the logged ones nothing but `status: "logged"` in `meta.json`, then one offer to start the first. A bug flow is where the gate earns its keep: the prevention actions `$flow-bug-postmortem` wrote, the neighbouring bugs `$flow-bug-validate` found and the similar-risk patterns `$flow-bug-fix` flagged all land here — half a dozen of them, from one fix, and «similar risk elsewhere» that no skeptic ever read is exactly the entry that turns a closed bug into six open questions. Repeat for anything still `proposed` on the re-run that confirms the merge.
+- **Specialists created — `$flow-feat-ship` §5.1, in full and unchanged.** Read `meta.json.specialists[]` (flow-core §6.6); nothing `useful` and pending → skip silently. Otherwise one batched offer to save each as an agent, after the MR/PR and after the knowledge `$flow-bug-postmortem` §4 already consolidated — asked in every mode, never overwriting an agent that exists.
 - **Conventions learned — `$flow-feat-ship` §6.4bis, in full and unchanged.** Read `meta.json.conventions_candidates[]` (flow-core §8); nothing pending → skip silently. Otherwise: drop what `FLOW.md` already says, then one batched `AskUserQuestion` (**Add to FLOW.md** / **Not a rule** / **Later**, or **Replace** when it contradicts a line that is there). Asked in every mode — it edits a file in the tree. A fix is where the corrections land hardest: *«never mock the repository in that layer»*, *«the regression test goes next to the unit test, not in the functional suite»* — said once inside `bug:fix`, and proposed again in the next incident unless they reach the file.
 - **Close the follow-up that started this work.** `meta.json.origin` set and this ship moved `phase` to `done` → open the originating work's `meta.json` (live folder or `_archive/`) and set that `followups[]` entry to `status: "done"`. Best-effort: the originating folder is gone or unreadable → warn in one line and continue, never block. Without this the entry stays `in_progress` forever and `status`/`daily` keep reporting work that is finished.
 - **Only once `phase` is `done`**: delete the candidate anchor `$flow-bug-review §1.5` left behind — `git update-ref -d refs/flow/candidate/<TICKET>`. No question: local, never pushed, and the gate that needed it has run.

package/adapters/codex/skills/flow-doctor/SKILL.md

@@ -196,8 +196,8 @@ refuse) · **degraded** (it runs, quietly worse than the config promises), plus
 
 - Every **role** set in `agents.*` (not `fanout_max` / `fanout_tool` — not agents), plus
   `quality.reviewers` and `quality.review_skill`: discoverable in `~/.claude/agents`, the repo's
-  `.agents/agents`, or a plugin. Report each missing name and that it falls back to `general-purpose`
-  (a skill: skipped) — a panel missing agents reviews less and still reports a clean pass.
+  `.agents/agents`, or a plugin. Report each missing name and that it falls back per flow-core §6.6 — the installed agent
+  that fits, else `general-purpose` (a skill: skipped) — a panel missing agents reviews less and still reports a clean pass.
 - Harness cannot set a model per subagent → note that every `models.*` value degrades to inheritance.
 - **Where the live panel will go this session** (flow-core §4): a tool whose name ends in `panel_set` / `panel_patch` — namespaced by the harness (`mcp__panel__panel_set` on Claude Code) and possibly **deferred**, which reads as absent until the list of deferred tools is checked → the terminal draws the panel and no `panel.json` is written; genuinely absent → the panel is that file, as before. Neither is a fault — one line naming the tools as they are actually called here, or saying the file is the transport.
 
@@ -235,7 +235,7 @@ refuse) · **degraded** (it runs, quietly worse than the config promises), plus
   a non-negative integer (`0` = no ceiling; else the default `12` applies). Report both effective
   numbers with the count of panel members `quality.review_skill`/`quality.reviewers` defines, so
   the widest command's cost is visible before it is paid, not after.
-- **A task-list tool** (`TaskCreate` on Claude Code, possibly deferred) is exposed. Absent → `build` and `fix` track their plan steps as a checklist in the phase artifact instead, and the terminal draws no step list — `warn`, not `fail`. On Claude Code the usual cause is a model the harness withholds it from by default: name the fix, `"CLAUDE_CODE_ENABLE_TODO_TOOLS": "true"` under `env` in `~/.claude/settings.json`.
+- **A task-list tool** (`TaskCreate` on Claude Code, possibly deferred) — informational only. The plan steps are shown in the live panel and recorded in the phase artifact; a harness task list is an extra. Absent on Claude Code → one line saying it is withheld from newer models unless `"CLAUDE_CODE_ENABLE_TODO_TOOLS": "true"` is set under `env`, for whoever wants it without a panel.
 - **The `flow-workflows` CLI, which `review` reaches for before it composes a panel**:
   `npx flow-workflows@<version> --version` answers, with `<version>` read from `version` in
   `~/.claude/flow/.claude-plugin/plugin.json` — the same pin `review` uses, so the number that

package/adapters/codex/skills/flow-feat-build/SKILL.md

@@ -139,8 +139,8 @@ Take the shape from the **Access paths** table in `03-design.md`; a query matchi
 
 Execution mode:
 
-- **Single-thread (XS/S/M)**: implement yourself, step by step; subagents only as point consultants when blocked — `agents.architecture` from `FLOW.md` for layer questions, `agents.persistence` for query/mapping questions (`Agent general-purpose` if either is empty).
-- **Partial delegation (M/L with clear pieces)**: `Agent` for isolated endpoints, plus `agents.testing` from `FLOW.md` in parallel to prepare the test suite (`Agent general-purpose` if empty). Pass the full `03-design.md` in the prompt so agents do not invent things.
+- **Single-thread (XS/S/M)**: implement yourself, step by step. Delegate a piece when a specialist would do it clearly better than you — a UI component to a frontend agent in a mostly backend work, a migration to a persistence agent — not because it is tedious; consult one when blocked — `agents.architecture` from `FLOW.md` for layer questions, `agents.persistence` for query/mapping questions. Which agent each role resolves to, and whether you may pick past a named one, is flow-core §6.6. A delegated piece follows the brief rules of the next bullet and the one-writer rule of flow-core §6.
+- **Partial delegation (M/L with clear pieces)**: `Agent` for isolated endpoints, plus `agents.testing` from `FLOW.md` in parallel to prepare the test suite (empty → flow-core §6.6). Pass the full `03-design.md` in the prompt so agents do not invent things.
 
   A brief here states the **output** as explicitly as the input (flow-core §6):
   - **Name the path the agent writes**, and require it to **save after each finished piece, not at the end**. Code, tests and generated content are the deliverable; the report is not. Then **verify the file** instead of believing the report — a piece half-written on disk is recoverable, what only exists in the agent's context is lost the moment it is stopped.
@@ -150,7 +150,9 @@ Execution mode:
 
 Use `TaskCreate` to track the steps of the **§2.0ter plan for this MR/PR** — one task per numbered step, same order. Step skipped there (small MR/PR) → seed from `03-design.md` §"Implementation plan (order)", taking only the steps this MR/PR covers per `04-mr-plan.md`; the feature's plan predates the split and carries steps that belong to siblings. Mark each step `in_progress` when starting and `completed` when done — do not batch.
 
-**Find the task-list tool before you conclude there is none.** `TaskCreate` is the Claude Code name; a harness may call it otherwise (`TodoWrite` on older builds), list it **deferred** — by name only, uncallable until its schema is loaded — or not offer it at all: Claude Code withholds it from newer models unless `CLAUDE_CODE_ENABLE_TODO_TOOLS=true` is set. A deferred tool is a tool that is there: load it and seed the list. Only when the search comes back empty, the plan in `05-implementation.md` **is** the list — tick each step there (`- [x]`) as it lands, same cadence, and say it once in the first step report (`no task-list tool here — steps tracked in 05-implementation.md`), so the user does not wait for a list that will never draw. An instruction to use a tool that is not there is otherwise skipped in silence, and the plan is written and never followed.
+**The steps go to the live panel** (flow-core §4.1 (2b), §4.4 (f)) — that is the list the user watches, whatever the harness. Publish the block when the first step starts, patch it as each one lands.
+
+**A harness task list is optional on top.** `TaskCreate` is the Claude Code name; a harness may call it otherwise (`TodoWrite` on older builds), list it **deferred** — by name only, uncallable until its schema is loaded — or not offer it at all (Claude Code withholds it from newer models unless `CLAUDE_CODE_ENABLE_TODO_TOOLS=true`). There → seed it too; not there → nothing is missing and nothing is said. Either way the plan in `05-implementation.md` is the record: tick each step there (`- [x]`) as it lands — the panel and any task list follow the ticks, never the other way round.
 
 ### 2.1bis The premise this change rests on (any size, XS included)
 
@@ -201,7 +203,7 @@ The step's changes are **always reported before anything is recorded**. Who deci
 
 **After completing each plan step**:
 
-1. **Mark the step `completed`** in the task list — or tick it in `05-implementation.md` where there is none (§2.1).
+1. **Tick the step** in `05-implementation.md`, patch the panel's step block (flow-core §4.4 (f)), and mark it `completed` in the harness task list if there is one (§2.1).
 2. **Report to the user** a step summary (≤ 5 lines):
    ```
    Step N done: <description>

package/adapters/codex/skills/flow-feat-design/SKILL.md

@@ -175,7 +175,7 @@ Save the result at the top of `03-design.md` under "## What already exists". Des
 ## 4. Work
 
 1. Load the project's skills (`FLOW.md` section `conventions`).
-2. Launch **in parallel** by feature/project type. Agent = `agents.<role>` from `FLOW.md`; empty → `Agent general-purpose` with the role in the prompt.
+2. Launch **in parallel** by feature/project type. Agent = `agents.<role>` from `FLOW.md`; empty → the installed agent that fits, else `Agent general-purpose` with the role in the prompt (flow-core §6.6).
 
 | When | Agent | Task |
 |---|---|---|

package/adapters/codex/skills/flow-feat-ship/SKILL.md

@@ -234,6 +234,29 @@ Only if `knowledge.read_staging` or `knowledge.save` is set; neither → skip wi
 3. **Combine staging + new findings** into a short list. Empty or only code-derivable → do not insist.
 4. 1+ relevant findings → ask the user whether to consolidate. Yes → run the consolidation of flow-core §0 (`read_staging`, then one decision and one `knowledge.save` per finding). No → do not insist.
 
+## 5.1 Specialists created (offer)
+
+Read `meta.json.specialists[]` (flow-core §6.6). **No entry with `useful: true` and `status` unset
+or `"later"` → skip this section and say nothing.** It comes after §5 on purpose: the MR/PR is open,
+the knowledge is saved, and what is left is whether an agent this work improvised deserves a name.
+
+1. **A harness with no file-defined agents** → the section does not exist; say so in one line if
+   there were useful specialists, so the brief is not mistaken for lost (it stays in `meta.json`).
+2. **One `AskUserQuestion`**, at most 4 entries, most useful first: its `name`, the role, the phase,
+   and in one line what it caught. Options: **Save as agent** · **Save and use it for `<role>`** ·
+   **No** · **Later**. The second option exists only when `agents.<role>` is empty.
+3. **Asked in every mode**, `auto` included: it writes to your agent configuration, outside the
+   tree, and that is never flow mechanics.
+4. **Save** → write the agent where and in the format your harness reads agents from (on Claude
+   Code, `~/.claude/agents/<name>.md`: frontmatter `name` and a one-line `description` of when to
+   use it, then the brief as the body — drop what only made sense for this ticket, keep the stack,
+   the checks and the typical mistakes). A file with that name already there → never overwrite:
+   suffix the name and say so. **Save and use it** → also write `agents.<role>: <name>` to the
+   active overlay (a harness-owned agent name, flow-core §0). Then `status: "saved"`, with the path.
+   **No** → `"declined"`, never offered again. **Later** → `"later"`, offered by the re-run of this
+   close.
+5. **One line on screen**: how many agents were saved and where.
+
 ## 5.5 A worktree does not own the tree its tools wrote into
 
 `meta.json.worktree` null → skip silently; there is nothing here that can go wrong.

package/adapters/codex/skills/flow-work-README/SKILL.md

@@ -136,6 +136,12 @@ This system **orchestrates** the project's existing sub-agents and skills (it do
       "evidence": "the stop it came from",
       "status": "proposed" | "added" | "declined", "note": "" }
   ],
+  "specialists": [
+    { "name": "react-form-reviewer", "role": "frontend", "phase": "review",
+      "brief": "the full expert brief it was launched with",
+      "useful": true | false | null,
+      "status": null | "saved" | "declined" | "later", "path": null }
+  ],
   "origin": { "work": "{PREFIX}XXXXX", "followup": "F1" },
   "started_at": "2026-05-11T10:00:00Z",
   "updated_at": "2026-05-11T11:30:00Z",
@@ -152,6 +158,7 @@ This system **orchestrates** the project's existing sub-agents and skills (it do
 - **`followups`**: work a phase deliberately did **not** do here — a neighbouring defect found while building, a piece put out of scope while planning, a risk recorded but not mitigated, an edge case left unchecked, a prevention action from a postmortem. Written by the phase that parks it (asks nothing) when it clears the bar of flow-core §7, triaged **once** at `ship`'s Close behind a skeptic and a ceiling on how many may become questions, published in the MR/PR description while still open, and surfaced by `status`/`daily`/`next` — including from `_archive/` — until `declined` or `done`. Shape and rules: flow-core §7. `[]` when the work deferred nothing.
 - **`defaults_used`**: an empty `FLOW.md` key a phase resolved with its default instead of asking — written in `guided`/`auto`, where asking would breach the never-a-question contract (flow-core §0). Not an error and not deferred work: a record, aggregated by `$flow-doctor` across every work including `_archive/`, so a key that keeps defaulting can be pinned once instead of decided every time. `[]` when nothing defaulted.
 - **`conventions_candidates`**: rules the user taught **this repo's flow** while correcting a phase — *«not a listener, a message handler»*, *«the test command is `make test-unit`»*. Written by the phase that heard one, asking nothing, only when it would apply to an unrelated ticket in this repo; read by `review`'s idiom audit in the same work; offered once at `ship`'s Close (**Add to FLOW.md** / **Not a rule** / **Later**, or **Replace**), in every mode, because it edits a file in the tree. Never domain knowledge — that filter is unchanged. Shape and rules: flow-core §8. `[]` when the work taught nothing.
+- **`specialists`**: agents a phase **created for this work** because nothing installed fit a role, under `agents.selection: open` (flow-core §6.6) — the brief it wrote and ran on `general-purpose`, and whether anything it returned was kept. Nothing is written to the harness's agent folders while the work runs; `$flow-feat-ship` §5.1 offers the useful ones for saving as real agents once the MR/PR is open and the knowledge is saved, and records where each went.
 - **`origin`**: set when this work *is* a follow-up someone accepted — which work raised it and which entry. Lets `start` carry the recorded "why" into `01-context.md` instead of re-deriving it, and lets that entry move to `done` when this work does. Absent for a work that came straight from a ticket.
 - Ticket format follows `tracker.prefix`; empty → free-form.
 

package/adapters/gemini/CORE.md

@@ -16,7 +16,7 @@ Every `/flow:*` command assumes these rules. They are stated once, here, so a co
 carries what is specific to its phase. Read this once per session; a command that says "load
 `flow-core`" means this file.
 
-**This file belongs to flow `0.76.0`.** Compare it once, at the start of the session, against
+**This file belongs to flow `0.78.0`.** Compare it once, at the start of the session, against
 `version` in `~/.claude/flow/.claude-plugin/plugin.json`. The two differing means the session
 is running a **mixture** — the commands from one copy of the plugin, these shared rules from another
 — which is exactly what happens when a branch or a release candidate is loaded over an installed
@@ -97,8 +97,7 @@ a command relies on is simply absent in that session, and nothing else will repo
      "general-purpose", "phase": "review" }`. Asking here would break the never-a-question contract
      (§2); recording keeps the decision readable afterwards, and `/flow:doctor` aggregates it.
 
-  **The list, and it is short on purpose**: `agents.<role>` when a panel is about to improvise that
-  role on an M/L or sensitive diff · `quality.review_depth` after a review's cost line went over
+  **The list, and it is short on purpose**: `quality.review_depth` after a review's cost line went over
   `agents.budget_max` · `data.volumes` or `data.explain_cmd` when the query duel came back
   schema-only · `quality.functional_check` when `validate` found a runnable app and no way to drive
   it · `tracker.start_cmd`/`done_cmd`/`abandon_cmd` the first time a ticket could have been
@@ -156,8 +155,8 @@ improvises, `models.workers` for the parallel fan-out rounds, and `models.superv
 bounded waits of §6.5 — the last two falling back to `models.agents`. Empty, absent, or no section
 → everything runs on the model you were launched with, and you say nothing.
 
-- Pass the resolved value to every subagent you launch, **except** an agent named in `agents.<role>`,
-  which keeps the model its own definition sets.
+- Pass the resolved value to every subagent you launch, **except** an agent named in `agents.<role>`
+  or picked from the installed ones (§6.6), which keeps the model its own definition sets.
 - **You cannot switch your own model**, so no key names the phases the main agent performs itself.
   A phase that wants another model says it in one line at the handoff (`this step reads better on
   <value>, you are on <current>` → `/model <value>`), records it in the artifact, and continues.
@@ -295,6 +294,9 @@ prose describing a file is otherwise obeyed to the letter.
     "",
     {"ref": "#1", "text": "batch read sources", "mark": "wait", "link": "https://gitlab.com/…/merge_requests/9977"},
     {"ref": "#2", "text": "per-event and per-recipient counters", "mark": "current"},
+    {"ref": "1–2", "text": "counter columns + migration · repository query", "mark": "done", "indent": 4},
+    {"ref": "3", "text": "per-recipient aggregation in the use case", "mark": "current", "indent": 4},
+    {"ref": "4", "text": "expose both counters on the endpoint", "mark": "pending", "indent": 4},
     {"ref": "#3–#6", "text": "channel map · use case · document detail · route", "mark": "pending"},
     "",
     {"ref": "Now", "text": "unit suite and the test agent over #2", "mark": "info"},
@@ -306,10 +308,10 @@ prose describing a file is otherwise obeyed to the letter.
 }
 ```
 
-- **`mark` says what a line is**; the reader draws it. `done` · `current` (at most one) · `pending` · `wait` (waiting on someone else: an open MR/PR, a user decision) · `block` · `info`. Marked lines form an aligned column: symbol, `ref`, text, link pinned right. Do not set `style` on a marked line — except `mark: "info"` + `style: ok|warn|error` when the colour *is* the information (a monitoring verdict).
+- **`mark` says what a line is**; the reader draws it. `done` · `current` (at most one per block) · `pending` · `wait` (waiting on someone else: an open MR/PR, a user decision) · `block` · `info`. Marked lines form an aligned column: symbol, `ref`, text, link pinned right. Do not set `style` on a marked line — except `mark: "info"` + `style: ok|warn|error` when the colour *is* the information (a monitoring verdict).
 - **`ref` need not be a number**: `#1`, `#3–#6`, `Now`, `Next`, `Decision`. Column width is per block; blocks are separated by blank lines — use them so the MR/PR train and the labels below do not drag each other wide.
 - **`link` is a field, never a URL inside `text`** — and **every line that names an MR/PR carries it**: the train entries, a blocker about somebody else's MR/PR, a `Now` or `Next` about the open one. The URL is in `meta.json.mrs[].url` or in whatever recorded that blocker; a bare `!10327` typed into `text` is a number the reader cannot make clickable, which is the whole point of the field. No URL recorded anywhere → the number in `text` and nothing else, and go and record it.
-- **Order:** (1) work title (`style: title`, no mark); (2) the MR/PR train, one entry per `meta.json.mrs[]` with `ref` `#n`, short title, real-state `mark`, `link` when there is a URL — not-started entries collapse into one `#a–#z` `pending` line; omit the block without `mrs`; (3) `Now` — what is running, the one fact `meta.json` cannot hold; (4) `Next`; (5) `Decision`, `mark: wait`, **only** when parked on the user — over the tools it travels with `attention: "wait"` (§4.2), and both are cleared together; (6) blockers, `mark: block` (a sibling repo with `contract_handoff: pending`, a red pipeline, an unmerged dependency).
+- **Order:** (1) work title (`style: title`, no mark); (2) the MR/PR train, one entry per `meta.json.mrs[]` with `ref` `#n`, short title, real-state `mark`, `link` when there is a URL — not-started entries collapse into one `#a–#z` `pending` line; omit the block without `mrs`; (2b) **the plan steps of the MR/PR being built**, right under its train entry — only while `build` or `fix` is working through a written plan (`/flow:feat:build §2.0ter`, `/flow:bug:fix §2.0bis`); one line per step, `ref` its number (`1`, `2`…), `indent: 4`, `key` `step-<n>`, `mark` `done` · `current` · `pending` from the plan's ticks in the phase artifact — never ahead of them; past three finished steps they collapse into one `done` line (`1–4`, `key` `steps-done`), and past three pending ones the tail collapses the same way (`key` `steps-pending`), so the block stays at five lines or fewer; drop the block when the MR/PR's build ends. This is the step list the user watches: a harness's own task list, where there is one, is a bonus on top — the panel is what every harness can feed; (3) `Now` — what is running, the one fact `meta.json` cannot hold; (4) `Next`; (5) `Decision`, `mark: wait`, **only** when parked on the user — over the tools it travels with `attention: "wait"` (§4.2), and both are cleared together; (6) blockers, `mark: block` (a sibling repo with `contract_handoff: pending`, a red pipeline, an unmerged dependency).
 - **Rules:** `phase` is the phase you are **running now** (not `meta.json.phase`, which advances only at Close). `header: true` means ticket, type, phase and age are drawn by the reader — do not repeat them in `lines`. Keep it under ~14 lines. Every fact from `meta.json` and the artifacts — an invented MR/PR state is worse than a blank panel. Write in the language of the work's artifacts.
 
 ### 4.2 Over the terminal's MCP — `panel_set` once, `panel_patch` from then on
@@ -384,6 +386,10 @@ that never asked, because they answered it and the screen says it did not land;
 nobody clears keeps that tab claiming them for ever. The same applies to a blocker that lifted and
 to a `wait` line whose MR/PR merged.
 
+**(f) Each plan step that lands** (the (2b) block): one patch, the moment its tick goes into the
+phase artifact — that step to `done`, the next to `current`, `Now` moved on. Not batched at the end
+of the MR/PR: a list that jumps from nothing to finished told the user nothing while it mattered.
+
 ### 4.5 `meta.json.cost[]` — what a phase measured, when it could
 
 A phase that can measure what it spent appends one entry: `{ "phase": "review", "at": "<ISO>",
@@ -548,6 +554,61 @@ moving.
 than to run — the brief, the round trip and the report are the whole of the saving. And anything you
 must *react to* mid-wait is not a wait at all: it is the work, and it stays here.
 
+### 6.6 — Which agent a role resolves to
+
+**The harness already tells you which agents are installed**, each with a line on what it is for,
+and choosing the one that fits a piece of work is something you do well without being told. A rule
+that answers every empty role with `general-purpose` throws that away: a repo with a frontend
+specialist installed and no `agents.frontend` got a generalist told to act like one. So wherever a
+command says *"`agents.<role>`, empty → `Agent general-purpose` with the role"*, it means this:
+
+1. **`agents.<role>` names an agent** → that agent, as always.
+2. **Empty** → the installed agent whose description fits the role **and this work's stack** best —
+   read the list your harness gives you, never a guess at a name. None fits clearly →
+   `general-purpose` with the role in the prompt, as before. An agent that fits the role but not the
+   stack (a PHP reviewer on a React diff) does not fit.
+
+**`agents.selection`** decides how far your judgement reaches:
+
+- **`configured`** (empty → this) — the two steps above. A role you named is never second-guessed.
+- **`open`** — a named role is a default, not an order: for a piece clearly outside what it
+  covers, or where an installed agent fits this piece plainly better, take that one. And a piece
+  **no role covers** — a UI component in a work whose roles are all backend — may go to the
+  specialist that does cover it. Every substitution is **one line in the phase artifact**: role,
+  the agent configured, the agent used, and why. Unrecorded, a panel that ran with other agents
+  reads exactly like the one you configured.
+
+  **Nothing installed fits → create the specialist for this work.** An installed agent is, in the
+  end, a brief with a name; what made it good is the brief. So instead of *"act as frontend"*, write
+  the one an expert would carry — the stack and its versions as this repo uses them, what to check
+  in this piece, the mistakes typical of it, and the report contract of this section — and launch
+  `general-purpose` with it. Nothing is written to the harness's agent folders: the specialist lives
+  for the brief, and it works on every harness, because every one has a general-purpose agent.
+  Record it in `meta.json.specialists[]`:
+
+  `{ "name": "react-form-reviewer", "role": "frontend", "phase": "review", "brief": "<the full text
+  you sent>", "useful": null, "status": null }`
+
+  Once its result is in, set `useful`: `true` when something it returned was kept — a finding
+  accepted, a piece merged into the diff — `false` otherwise. Only a useful one is ever offered for
+  saving, and only **when the work ends** (`/flow:feat:ship` §5.1), never mid-phase: by then the
+  MR/PR is open and it is plain which of them earned a place among your agents.
+
+Three things this never touches, in either mode:
+
+- **The review roster** (`quality.review_skill` / `quality.reviewers`) runs as defined, with no
+  substitutions: its value is being the same panel every time, so a result can be compared.
+- **The agents that are `general-purpose` by design** — challengers, skeptics, the completeness
+  critic, the blinded contract check. They are generalists *on purpose*: a specialist brings the
+  assumptions the blinding is there to strip. A step that writes `general-purpose` with no role
+  behind it means exactly that.
+- **The budget.** A picked agent counts against `agents.budget_max` like any other. Like a named
+  agent it keeps its own model (§1), and `models.agents` reaches it only when it has none.
+
+Record what you picked where a default would be recorded (§0): *"`agents.frontend` empty → used
+`react-expert` (installed)"*. That line is also the evidence for writing the role down when the
+next stop offers it.
+
 ## 7. Deferred work — `followups[]`
 
 Every chain in this plugin reaches moments where the right call is **not to do it here**: a build

package/adapters/gemini/commands/flow/bug/fix.toml

@@ -91,7 +91,7 @@ Save it in `04-fix.md` under `## Plan`, right after the brief. **A fix that was
 
 - Apply the minimal fix targeting the finding of `03-investigation.md` (or the diagnosis if investigate was skipped).
 - Sensitive area (authentication, payments, sensitive data) → consult the `agents.architecture` agent from FLOW.md to confirm the correct layer; empty → check against `conventions` in FLOW.md.
-- More than 2 fix steps → the §2.0bis plan is already written: track its numbered steps with `TaskCreate` where the harness offers one — looked for as `/flow:feat:build §2.1` says, since it may be deferred or named otherwise — otherwise tick them off in `04-fix.md` where they stand and say so once in the first step report. One task per step, same order, marked `in_progress` and `completed` as they go — not batched.
+- More than 2 fix steps → the §2.0bis plan is already written: tick its numbered steps off in `04-fix.md` where they stand, and show them in the live panel's step block (flow-core §4.1 (2b), patched per §4.4 (f)); a harness task list, if there is one, goes on top as `/flow:feat:build §2.1` says. One task per step, same order, marked `in_progress` and `completed` as they go — not batched.
 - **Comment discipline**: comment only a *why* the code cannot say (a non-obvious constraint, the reason for a workaround, a subtle invariant); match the surrounding file's comment density. **Never write the ticket ID or "fix for #N" into a code comment** — that lives in the commit/branch/MR-PR.
 - Keep the log updated while editing.
 

package/adapters/gemini/commands/flow/bug/ship.toml

@@ -182,6 +182,7 @@ nothing was measured. A slowness bug is the case where this comment *is* the pro
 - **Cross-repo reminder**: for each `meta.json.related_repos` entry not `done`: *"you've shipped the `<this-repo>` part; `<repo>` still needs: `<scope>` → go there and run `/flow:bug:start <TICKET>` (or `/flow:feat:start`)"*. flow only reminds; this is not a hard gate.
 - **Cross-repo contract handoff**: for each such entry with `contract_handoff: "pending"` — the fix changed a surface that sibling consumes — hand the new shape over as `/flow:feat:ship` §6.3 (publish the **literal** shape to the tracker ticket, mandatory preview in every autonomy mode, then record `published → <location>`; versioned-file fallback when there is no tracker). A bug flow has no §"External contracts": take the literal from the fix itself — the actual key, code, route or event the code now emits — read it off the code, never from memory.
 - **Follow-up triage — `/flow:feat:ship` §6.4, in full and unchanged.** Read `meta.json.followups[]`; nothing `proposed` → skip silently. Otherwise §6.4 step 0 first — **3 or more candidates go to a skeptic** whose job is to refute that each deserves a ticket, and only the strongest `tracker.followup_ask_max` (empty → **2**) reach the user (flow-core §7) — then one batched `AskUserQuestion` (**Do it** / **Not worth it** / **Later**; `decision` entries asked as their question, `tooling` and everything `log`ged never asked), the accepted ones get their tracker issue created (outward-facing → **asks in every mode**), the logged ones nothing but `status: "logged"` in `meta.json`, then one offer to start the first. A bug flow is where the gate earns its keep: the prevention actions `/flow:bug:postmortem` wrote, the neighbouring bugs `/flow:bug:validate` found and the similar-risk patterns `/flow:bug:fix` flagged all land here — half a dozen of them, from one fix, and «similar risk elsewhere» that no skeptic ever read is exactly the entry that turns a closed bug into six open questions. Repeat for anything still `proposed` on the re-run that confirms the merge.
+- **Specialists created — `/flow:feat:ship` §5.1, in full and unchanged.** Read `meta.json.specialists[]` (flow-core §6.6); nothing `useful` and pending → skip silently. Otherwise one batched offer to save each as an agent, after the MR/PR and after the knowledge `/flow:bug:postmortem` §4 already consolidated — asked in every mode, never overwriting an agent that exists.
 - **Conventions learned — `/flow:feat:ship` §6.4bis, in full and unchanged.** Read `meta.json.conventions_candidates[]` (flow-core §8); nothing pending → skip silently. Otherwise: drop what `FLOW.md` already says, then one batched `AskUserQuestion` (**Add to FLOW.md** / **Not a rule** / **Later**, or **Replace** when it contradicts a line that is there). Asked in every mode — it edits a file in the tree. A fix is where the corrections land hardest: *«never mock the repository in that layer»*, *«the regression test goes next to the unit test, not in the functional suite»* — said once inside `bug:fix`, and proposed again in the next incident unless they reach the file.
 - **Close the follow-up that started this work.** `meta.json.origin` set and this ship moved `phase` to `done` → open the originating work's `meta.json` (live folder or `_archive/`) and set that `followups[]` entry to `status: "done"`. Best-effort: the originating folder is gone or unreadable → warn in one line and continue, never block. Without this the entry stays `in_progress` forever and `status`/`daily` keep reporting work that is finished.
 - **Only once `phase` is `done`**: delete the candidate anchor `/flow:bug:review §1.5` left behind — `git update-ref -d refs/flow/candidate/<TICKET>`. No question: local, never pushed, and the gate that needed it has run.

package/adapters/gemini/commands/flow/doctor.toml

@@ -192,8 +192,8 @@ refuse) · **degraded** (it runs, quietly worse than the config promises), plus
 
 - Every **role** set in `agents.*` (not `fanout_max` / `fanout_tool` — not agents), plus
   `quality.reviewers` and `quality.review_skill`: discoverable in `~/.claude/agents`, the repo's
-  `.agents/agents`, or a plugin. Report each missing name and that it falls back to `general-purpose`
-  (a skill: skipped) — a panel missing agents reviews less and still reports a clean pass.
+  `.agents/agents`, or a plugin. Report each missing name and that it falls back per flow-core §6.6 — the installed agent
+  that fits, else `general-purpose` (a skill: skipped) — a panel missing agents reviews less and still reports a clean pass.
 - Harness cannot set a model per subagent → note that every `models.*` value degrades to inheritance.
 - **Where the live panel will go this session** (flow-core §4): a tool whose name ends in `panel_set` / `panel_patch` — namespaced by the harness (`mcp__panel__panel_set` on Claude Code) and possibly **deferred**, which reads as absent until the list of deferred tools is checked → the terminal draws the panel and no `panel.json` is written; genuinely absent → the panel is that file, as before. Neither is a fault — one line naming the tools as they are actually called here, or saying the file is the transport.
 
@@ -231,7 +231,7 @@ refuse) · **degraded** (it runs, quietly worse than the config promises), plus
   a non-negative integer (`0` = no ceiling; else the default `12` applies). Report both effective
   numbers with the count of panel members `quality.review_skill`/`quality.reviewers` defines, so
   the widest command's cost is visible before it is paid, not after.
-- **A task-list tool** (`TaskCreate` on Claude Code, possibly deferred) is exposed. Absent → `build` and `fix` track their plan steps as a checklist in the phase artifact instead, and the terminal draws no step list — `warn`, not `fail`. On Claude Code the usual cause is a model the harness withholds it from by default: name the fix, `"CLAUDE_CODE_ENABLE_TODO_TOOLS": "true"` under `env` in `~/.claude/settings.json`.
+- **A task-list tool** (`TaskCreate` on Claude Code, possibly deferred) — informational only. The plan steps are shown in the live panel and recorded in the phase artifact; a harness task list is an extra. Absent on Claude Code → one line saying it is withheld from newer models unless `"CLAUDE_CODE_ENABLE_TODO_TOOLS": "true"` is set under `env`, for whoever wants it without a panel.
 - **The `flow-workflows` CLI, which `review` reaches for before it composes a panel**:
   `npx flow-workflows@<version> --version` answers, with `<version>` read from `version` in
   `~/.claude/flow/.claude-plugin/plugin.json` — the same pin `review` uses, so the number that

package/adapters/gemini/commands/flow/feat/build.toml

@@ -135,8 +135,8 @@ Take the shape from the **Access paths** table in `03-design.md`; a query matchi
 
 Execution mode:
 
-- **Single-thread (XS/S/M)**: implement yourself, step by step; subagents only as point consultants when blocked — `agents.architecture` from `FLOW.md` for layer questions, `agents.persistence` for query/mapping questions (`Agent general-purpose` if either is empty).
-- **Partial delegation (M/L with clear pieces)**: `Agent` for isolated endpoints, plus `agents.testing` from `FLOW.md` in parallel to prepare the test suite (`Agent general-purpose` if empty). Pass the full `03-design.md` in the prompt so agents do not invent things.
+- **Single-thread (XS/S/M)**: implement yourself, step by step. Delegate a piece when a specialist would do it clearly better than you — a UI component to a frontend agent in a mostly backend work, a migration to a persistence agent — not because it is tedious; consult one when blocked — `agents.architecture` from `FLOW.md` for layer questions, `agents.persistence` for query/mapping questions. Which agent each role resolves to, and whether you may pick past a named one, is flow-core §6.6. A delegated piece follows the brief rules of the next bullet and the one-writer rule of flow-core §6.
+- **Partial delegation (M/L with clear pieces)**: `Agent` for isolated endpoints, plus `agents.testing` from `FLOW.md` in parallel to prepare the test suite (empty → flow-core §6.6). Pass the full `03-design.md` in the prompt so agents do not invent things.
 
   A brief here states the **output** as explicitly as the input (flow-core §6):
   - **Name the path the agent writes**, and require it to **save after each finished piece, not at the end**. Code, tests and generated content are the deliverable; the report is not. Then **verify the file** instead of believing the report — a piece half-written on disk is recoverable, what only exists in the agent's context is lost the moment it is stopped.
@@ -146,7 +146,9 @@ Execution mode:
 
 Use `TaskCreate` to track the steps of the **§2.0ter plan for this MR/PR** — one task per numbered step, same order. Step skipped there (small MR/PR) → seed from `03-design.md` §"Implementation plan (order)", taking only the steps this MR/PR covers per `04-mr-plan.md`; the feature's plan predates the split and carries steps that belong to siblings. Mark each step `in_progress` when starting and `completed` when done — do not batch.
 
-**Find the task-list tool before you conclude there is none.** `TaskCreate` is the Claude Code name; a harness may call it otherwise (`TodoWrite` on older builds), list it **deferred** — by name only, uncallable until its schema is loaded — or not offer it at all: Claude Code withholds it from newer models unless `CLAUDE_CODE_ENABLE_TODO_TOOLS=true` is set. A deferred tool is a tool that is there: load it and seed the list. Only when the search comes back empty, the plan in `05-implementation.md` **is** the list — tick each step there (`- [x]`) as it lands, same cadence, and say it once in the first step report (`no task-list tool here — steps tracked in 05-implementation.md`), so the user does not wait for a list that will never draw. An instruction to use a tool that is not there is otherwise skipped in silence, and the plan is written and never followed.
+**The steps go to the live panel** (flow-core §4.1 (2b), §4.4 (f)) — that is the list the user watches, whatever the harness. Publish the block when the first step starts, patch it as each one lands.
+
+**A harness task list is optional on top.** `TaskCreate` is the Claude Code name; a harness may call it otherwise (`TodoWrite` on older builds), list it **deferred** — by name only, uncallable until its schema is loaded — or not offer it at all (Claude Code withholds it from newer models unless `CLAUDE_CODE_ENABLE_TODO_TOOLS=true`). There → seed it too; not there → nothing is missing and nothing is said. Either way the plan in `05-implementation.md` is the record: tick each step there (`- [x]`) as it lands — the panel and any task list follow the ticks, never the other way round.
 
 ### 2.1bis The premise this change rests on (any size, XS included)
 
@@ -197,7 +199,7 @@ The step's changes are **always reported before anything is recorded**. Who deci
 
 **After completing each plan step**:
 
-1. **Mark the step `completed`** in the task list — or tick it in `05-implementation.md` where there is none (§2.1).
+1. **Tick the step** in `05-implementation.md`, patch the panel's step block (flow-core §4.4 (f)), and mark it `completed` in the harness task list if there is one (§2.1).
 2. **Report to the user** a step summary (≤ 5 lines):
    ```
    Step N done: <description>

package/adapters/gemini/commands/flow/feat/design.toml

@@ -171,7 +171,7 @@ Save the result at the top of `03-design.md` under "## What already exists". Des
 ## 4. Work
 
 1. Load the project's skills (`FLOW.md` section `conventions`).
-2. Launch **in parallel** by feature/project type. Agent = `agents.<role>` from `FLOW.md`; empty → `Agent general-purpose` with the role in the prompt.
+2. Launch **in parallel** by feature/project type. Agent = `agents.<role>` from `FLOW.md`; empty → the installed agent that fits, else `Agent general-purpose` with the role in the prompt (flow-core §6.6).
 
 | When | Agent | Task |
 |---|---|---|

package/adapters/gemini/commands/flow/feat/ship.toml

@@ -230,6 +230,29 @@ Only if `knowledge.read_staging` or `knowledge.save` is set; neither → skip wi
 3. **Combine staging + new findings** into a short list. Empty or only code-derivable → do not insist.
 4. 1+ relevant findings → ask the user whether to consolidate. Yes → run the consolidation of flow-core §0 (`read_staging`, then one decision and one `knowledge.save` per finding). No → do not insist.
 
+## 5.1 Specialists created (offer)
+
+Read `meta.json.specialists[]` (flow-core §6.6). **No entry with `useful: true` and `status` unset
+or `"later"` → skip this section and say nothing.** It comes after §5 on purpose: the MR/PR is open,
+the knowledge is saved, and what is left is whether an agent this work improvised deserves a name.
+
+1. **A harness with no file-defined agents** → the section does not exist; say so in one line if
+   there were useful specialists, so the brief is not mistaken for lost (it stays in `meta.json`).
+2. **One `AskUserQuestion`**, at most 4 entries, most useful first: its `name`, the role, the phase,
+   and in one line what it caught. Options: **Save as agent** · **Save and use it for `<role>`** ·
+   **No** · **Later**. The second option exists only when `agents.<role>` is empty.
+3. **Asked in every mode**, `auto` included: it writes to your agent configuration, outside the
+   tree, and that is never flow mechanics.
+4. **Save** → write the agent where and in the format your harness reads agents from (on Claude
+   Code, `~/.claude/agents/<name>.md`: frontmatter `name` and a one-line `description` of when to
+   use it, then the brief as the body — drop what only made sense for this ticket, keep the stack,
+   the checks and the typical mistakes). A file with that name already there → never overwrite:
+   suffix the name and say so. **Save and use it** → also write `agents.<role>: <name>` to the
+   active overlay (a harness-owned agent name, flow-core §0). Then `status: "saved"`, with the path.
+   **No** → `"declined"`, never offered again. **Later** → `"later"`, offered by the re-run of this
+   close.
+5. **One line on screen**: how many agents were saved and where.
+
 ## 5.5 A worktree does not own the tree its tools wrote into
 
 `meta.json.worktree` null → skip silently; there is nothing here that can go wrong.

package/adapters/gemini/commands/flow/work/README.toml

@@ -132,6 +132,12 @@ This system **orchestrates** the project's existing sub-agents and skills (it do
       "evidence": "the stop it came from",
       "status": "proposed" | "added" | "declined", "note": "" }
   ],
+  "specialists": [
+    { "name": "react-form-reviewer", "role": "frontend", "phase": "review",
+      "brief": "the full expert brief it was launched with",
+      "useful": true | false | null,
+      "status": null | "saved" | "declined" | "later", "path": null }
+  ],
   "origin": { "work": "{PREFIX}XXXXX", "followup": "F1" },
   "started_at": "2026-05-11T10:00:00Z",
   "updated_at": "2026-05-11T11:30:00Z",
@@ -148,6 +154,7 @@ This system **orchestrates** the project's existing sub-agents and skills (it do
 - **`followups`**: work a phase deliberately did **not** do here — a neighbouring defect found while building, a piece put out of scope while planning, a risk recorded but not mitigated, an edge case left unchecked, a prevention action from a postmortem. Written by the phase that parks it (asks nothing) when it clears the bar of flow-core §7, triaged **once** at `ship`'s Close behind a skeptic and a ceiling on how many may become questions, published in the MR/PR description while still open, and surfaced by `status`/`daily`/`next` — including from `_archive/` — until `declined` or `done`. Shape and rules: flow-core §7. `[]` when the work deferred nothing.
 - **`defaults_used`**: an empty `FLOW.md` key a phase resolved with its default instead of asking — written in `guided`/`auto`, where asking would breach the never-a-question contract (flow-core §0). Not an error and not deferred work: a record, aggregated by `/flow:doctor` across every work including `_archive/`, so a key that keeps defaulting can be pinned once instead of decided every time. `[]` when nothing defaulted.
 - **`conventions_candidates`**: rules the user taught **this repo's flow** while correcting a phase — *«not a listener, a message handler»*, *«the test command is `make test-unit`»*. Written by the phase that heard one, asking nothing, only when it would apply to an unrelated ticket in this repo; read by `review`'s idiom audit in the same work; offered once at `ship`'s Close (**Add to FLOW.md** / **Not a rule** / **Later**, or **Replace**), in every mode, because it edits a file in the tree. Never domain knowledge — that filter is unchanged. Shape and rules: flow-core §8. `[]` when the work taught nothing.
+- **`specialists`**: agents a phase **created for this work** because nothing installed fit a role, under `agents.selection: open` (flow-core §6.6) — the brief it wrote and ran on `general-purpose`, and whether anything it returned was kept. Nothing is written to the harness's agent folders while the work runs; `/flow:feat:ship` §5.1 offers the useful ones for saving as real agents once the MR/PR is open and the knowledge is saved, and records where each went.
 - **`origin`**: set when this work *is* a follow-up someone accepted — which work raised it and which entry. Lets `start` carry the recorded "why" into `01-context.md` instead of re-deriving it, and lets that entry move to `done` when this work does. Absent for a work that came straight from a ticket.
 - Ticket format follows `tracker.prefix`; empty → free-form.