@jentrix/plugin-claude 0.6.0 → 0.6.2

package/.claude-plugin/plugin.json

@@ -1,5 +1,5 @@
 {
   "name": "jentrix",
   "description": "Connect this Claude Code session to a Jentrix project: /jentrix-connect binds the current session (trusted lifecycle hooks supply the session id), /jentrix-align anchors it to a task with an accountable owner, /jentrix-plan, /jentrix-checkpoint, /jentrix-review, /jentrix-status and /jentrix-end wrap the same session operations. Business logic lives in the Jentrix CLI — removing this plugin leaves `jentrix session connect` as the universal fallback.",
-  "version": "0.6.0"
+  "version": "0.6.2"
 }

package/commands/jentrix-align.md

@@ -47,6 +47,26 @@ wizard; a Project is an optional task label and is never part of it).
    `/jentrix-end` moves it to In review when the work is genuinely finished —
    the terminal move is the operator's Accept, never yours.
 
+5. **Finish by reading the card**, and relay what comes back:
+
+   ```
+   jentrix task context --task <that task>
+   ```
+
+   One call returns the description, the links BOTH ways with their titles,
+   the subtasks, the latest artifact of each type with its id, the last five
+   comments, and **Related evidence** — what OTHER work in this workspace
+   already said about this card (learnings, decisions, findings, reports
+   ranked by similarity; the align output printed the same block). Read it
+   before you touch any code: a linked card is usually where the prior
+   round's reasoning lives, its artifacts are the context you would otherwise
+   reconstruct from scratch, and a related learning is a mistake you do not
+   have to make twice — `jentrix artifact get --artifact-id <id>` reads one in
+   full. If it prints `no links · no artifacts · no comments · no related
+   evidence`, the card has nothing to give and you stop looking — that single
+   line is the answer, not an invitation to run `task get`, `task search` and
+   `comment list` to confirm it.
+
 Notes for you (the assistant):
 
 - Do NOT pass `--provider-session` — the CLI reads the current session id and

package/commands/jentrix-checkpoint.md

@@ -45,27 +45,63 @@ the plan, when a long investigation resolves, or before handing the work over.
 2. Do not summarize the transcript. Do not restate the prompt. Do not include
    anything you have not verified this session — a checkpoint that carries a
    guess as a fact poisons every session that reads it.
-3. Push it:
+3. **If the checkpoint claims a gate, run the gate first.** A checkpoint that
+   says "typecheck green" without an attested LOG behind it is a claim, and
+   `jentrix session end` now REFUSES a close whose report claims a gate family
+   (`typecheck` · `lint` · `test` · `e2e`) with no attested LOG of that same
+   family. So, before writing the body: for each gate you are about to name,
+   check whether this session already has an attested LOG for it, and if not,
+   run it —
+
+   ```
+   jentrix push log --from-cmd "<the gate command>"
+   ```
+
+   — then cite the returned artifact id in the checkpoint beside the claim.
+   The command exits with the GATE's own code, so a red gate stays red and you
+   report it red. If you are not going to run it, do not claim it: say what was
+   not run and why.
+
+4. **`## Code contact`** — the files the objective named, and which of them
+   this session actually opened:
+
+   ```
+   jentrix session contact --paths <the files the PROMPT/GOAL named>
+   ```
+
+   List the never-opened ones under that heading, verbatim. This is the check
+   that catches the expensive failure: work built on a file nobody read. If the
+   answer says no local activity skeleton exists, say THAT — "never opened" by
+   absence is not the same claim as "never opened" by evidence.
+
+5. Push it — as an ARTIFACT, never as a comment. A "Checkpoint" posted as a
+   comment is not a checkpoint: the summary counts REPORTs titled
+   `Checkpoint — …` and prints `checkpoints: N`, and a comment counts zero.
    - `jentrix push learning --title "Checkpoint — <topic>"` when it is durable
      knowledge worth carrying past this work;
    - `jentrix push report --title "Checkpoint — <topic>"` when it is the
      state of THIS work.
    Content goes on stdin.
-4. Relay the returned `artifactId` to the operator, and say which kind you
+6. Relay the returned `artifactId` to the operator, and say which kind you
    chose and why.
-5. If the session's task status no longer matches reality, fix it now:
+7. If the session's task status no longer matches reality, fix it now:
    `jentrix column list --board <boardId>` then
    `jentrix task move --task <id> --to-column-id <id>`. Only the columns the
    board actually has.
 
 **Cadence duty (evidence floor).** Checkpoint after EVERY commit, not only at
-the end: push the `decision` (with `--basis`) the commit rested on the moment
-it lands, and run its gates through `jentrix push log --from-cmd "<command>"`
-so the exit code and output tail are recorded as a LOG. Record quality must
-track the work, not end-of-session diligence — `jentrix session end` enforces
-this (it refuses on commits with no decision record and on sectioned reports
-with no typed artifacts; a deliberate exception is an ordinary
-`jentrix push gap`).
+the end: push the `decision` the commit rested on the moment it lands — with
+`--basis <artifact-id-or-url>`, or `--no-basis "<reason>"` when it rested on
+your own reading and there is nothing filed to cite (the command REFUSES
+without one of the two) — and run its gates through `jentrix push log
+--from-cmd "<command>"` so the exit code and output tail are recorded as a LOG.
+Record quality must track the work, not end-of-session diligence — `jentrix
+session end` enforces this: it refuses on commits with no decision record that
+names its basis, on sectioned reports with no typed artifacts, and on a gate
+claimed in prose with no attested LOG of that gate family. A deliberate
+exception is an ordinary `jentrix push gap` — but a gap that names a file in
+the checkout this session never opened is itself refused, because that is not a
+limitation, it is an unread file.
 
 Note for the operator: this records the distillation. On a session aligned
 with `--capture`, the raw pre-compaction range is preserved separately and

package/commands/jentrix-end.md

@@ -20,15 +20,26 @@ pipeline), the typed artifacts ARE the record — push them before closing:
    - Push the final report: write a concise session report (what was done,
      what changed, what's next) and run
      `jentrix push report --title "<short title>"` with the content on stdin.
-   - Push any decisions made this session that were not already pushed:
-     `jentrix push decision --basis <artifact-id|url> …` (one per decision;
-     `--basis` is repeatable and names what the decision rested on — file the
-     context first so the ref resolves).
+   - Push any decisions made this session that were not already pushed. **Ask
+     what the decision rested on BEFORE you write the memo** — the basis is
+     not a field to fill in afterwards, it is the thing that makes the memo a
+     decision rather than an assertion. For each one, answer "what did this
+     rest on?" first, then push:
+     `jentrix push decision --basis <artifact-id|url> …` (repeatable; file the
+     context first so the ref resolves), or, when it genuinely rested on your
+     own reading of the code and there is nothing filed to cite,
+     `jentrix push decision --no-basis "<the reason>"`. The command REFUSES
+     without one of the two, and a memo with neither would not have counted
+     toward E2/E3 anyway.
    - **Gap sweep**: every promised-but-undone item becomes a
      `jentrix push gap` — one per gap, title stating the claim ("Windows hook
      path untested"). A session with no output MUST have at least one gap
      explaining why. Push `jentrix push issue` for anything found broken and
-     not yet recorded.
+     not yet recorded. A gap that names a file in this checkout the session
+     never OPENED is refused with the path and the remedy: read it or drop the
+     path. That refusal is not an obstacle to route around — the one-line read
+     that satisfies it is the read that would have told you whether the gap is
+     real.
    - Push any durable lessons: `jentrix push learning …`.
    - Do NOT push a diff by hand. When HEAD moved, `jentrix session end`
      generates the real `git log --patch` for the session's range itself and
@@ -82,18 +93,28 @@ pipeline), the typed artifacts ARE the record — push them before closing:
    say so if the user seems to expect it.
 
 **Evidence floor.** `jentrix session end` enforces fixed checks and will
-REFUSE to close when they are unmet: E1 — HEAD moved but no attested DIFF
-(the CLI pushes the real patch itself; just re-run `session end` from the
-aligned checkout); E2 — commits with no `jentrix push decision` and no
-declared-deviation `jentrix push gap`; E3 — a REPORT carrying `## Learnings`
-/ `## Gaps` / `## Decisions` sections while the session has zero
-corresponding typed artifacts (split them into `push learning|gap|decision`,
-never one blob). Comply by pushing the named evidence and retrying, or
-deviate honestly with a `gap` — `--acknowledge-evidence-gaps` closes anyway
-and stamps each unmet check MISSING into Review readiness. Run every gate
-through `jentrix push log --from-cmd "<command>"` (it records the exit code
-and output tail as a LOG and exits with the command's own code), so "tests
-green" claims carry evidence instead of tripping the E4 advisory.
+REFUSE to close when they are unmet:
+
+- **E1** — HEAD moved but no attested DIFF (the CLI pushes the real patch
+  itself; just re-run `session end` from the aligned checkout).
+- **E2** — commits with no `jentrix push decision` **that names its basis**,
+  and no declared-deviation `jentrix push gap`. Basis-less memos do not count,
+  however many are pushed: the count was never the evidence.
+- **E3** — a REPORT carrying `## Learnings` / `## Gaps` / `## Decisions`
+  sections while the session has zero corresponding typed artifacts (split
+  them into `push learning|gap|decision`, never one blob).
+- **E4** — a REPORT that CLAIMS a gate refuses the close unless the session
+  has an attested LOG (`jentrix push log --from-cmd`) whose command names the
+  **same gate family**: `typecheck` · `lint` · `test` · `e2e`. Family, not
+  exact command — "tests green" is covered by `pnpm test:mvp` and is NOT
+  covered by `pnpm lint`. A pasted LOG covers nothing; the CLI warns
+  `UNATTESTED LOG` when you push one, and the summary marks it.
+
+Comply by pushing the named evidence and retrying, or deviate honestly with a
+`gap`. `--acknowledge-evidence-gaps` closes anyway and stamps each unmet check
+MISSING into Review readiness — reach for it when the gap is real, never to get
+past a check you could satisfy by running the gate. A refusal leaves the
+session AND its capture host running, so the comply work is still recorded.
 
 ## Shared boundaries
 

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@jentrix/plugin-claude",
-  "version": "0.6.0",
+  "version": "0.6.2",
   "description": "The official Jentrix plugin for Claude Code: connected-session commands and trusted lifecycle hooks, materialized by the Jentrix CLI (jentrix plugin install claude).",
   "keywords": [
     "jentrix",