@entelligentsia/forgecli 1.0.25 → 1.0.40

package/dist/forge-payload/commands/init.md

@@ -1,26 +1,17 @@
 ---
 name: init
-description: Use when the current project has no Forge SDLC instance and you need to bootstrap one from scratch. Use --migrate to migrate an existing store to Forge format (replaces the removed /forge:migrate command).
+description: Bootstrap Forge's LLM half — KB discovery, config generation, and artifact registration. Run once after `4ge init claude .` has scaffolded the project structure.
 ---
 
 # /forge:init
 
-You are the Forge init orchestrator. Your job is to generate a complete, project-specific
-AI software development lifecycle for the codebase in the current working directory.
+You are the Forge init orchestrator. Your job is to run the LLM half of the
+Forge bootstrap — knowledge-base discovery, config generation, and artifact
+registration — in the current working directory. The CLI half (`4ge init claude .`)
+has already scaffolded the project structure, installed the wfl-init.js driver,
+and installed this command wrapper.
 
-## Locate the Forge plugin
-
-Set `$FORGE_ROOT` to the plugin root provided by Claude Code:
-
-```
-FORGE_ROOT: !`echo "${CLAUDE_PLUGIN_ROOT}"`
-```
-
-`$FORGE_ROOT` is the directory containing `meta/`, `init/`, `hooks/`, and `commands/`.
-
-## Execute
-
-### Resume Detection
+## Resume Detection
 
 Before showing the pre-flight plan, check for an existing checkpoint:
 
@@ -36,7 +27,8 @@ If the file exists and contains valid JSON, inspect it:
   ```sh
   rm -f .forge/init-progress.json
   ```
-  Then proceed to the **Pre-flight Plan** (new 4-phase flow, no mode prompt).
+  Then proceed to the **Pre-flight interactive prompts** (new 4-phase flow, no mode prompt).
+  Set `startPhase = 1`.
 
 - If the file contains valid JSON with `lastPhase` in range 1–4 and a `timestamp`
   field, emit the resume banner:
@@ -55,105 +47,232 @@ Use the following mapping:
 | 2         | Phase 3                 |
 | 3         | Phase 4                 |
 
-If the user chooses to resume: jump to the mapped resume phase.
-
+If the user chooses to resume: set `startPhase = nextPhase`.
 If the user chooses to start over: delete `.forge/init-progress.json`
-(`rm -f .forge/init-progress.json`) and proceed to the **Pre-flight Plan**.
+(`rm -f .forge/init-progress.json`) and set `startPhase = 1`.
 
 If the file does not exist, or contains invalid JSON, or contains an
 unrecognised `lastPhase` value outside 1–4: delete any corrupt file and
-proceed to the **Pre-flight Plan**.
+set `startPhase = 1`.
 
 If parsing the file throws (malformed JSON): log a one-line warning
 `△ init-progress.json is malformed — deleting and starting fresh.`, delete
-the file, and proceed to the **Pre-flight Plan**.
+the file, and set `startPhase = 1`.
 
-### Hero
+## Hero
 
 Render the Forge hero block once per session:
 
 ```sh
-node "$FORGE_ROOT/tools/banners.cjs" forge
-node "$FORGE_ROOT/tools/banners.cjs" --subtitle "AI SDLC bootstrapper · forge:init v$(node -p "require('$FORGE_ROOT/.claude-plugin/plugin.json').version")"
+node .forge/tools/banners.cjs forge
+node .forge/tools/banners.cjs --subtitle "AI SDLC bootstrapper · forge:init"
 ```
 
 The hero runs once. If the user resumes mid-init, do NOT re-render the hero —
 just emit the phase banner for the resume target phase.
 
-### Flag handling
+## Flag Handling
 
 `--fast` and `--full` are accepted as no-ops for backwards compatibility with
-scripts and CI pipelines. Both flags proceed with the standard 4-phase base-pack
-init. The fast/full distinction was removed in v0.40.0.
+scripts and CI pipelines. Both flags proceed with the standard 4-phase flow.
+
+**`--migrate` flag:** When `$ARGUMENTS` contains `--migrate`, redirect to the
+store migration workflow. Read `.forge/config.json` to locate `paths.forgeRoot`,
+then read `{forgeRoot}/meta/workflows/meta-migrate.md` and follow it.
+Do NOT proceed to the pre-flight prompts when `--migrate` is present.
+
+## Pre-flight Interactive Prompts
+
+Gather these values before dispatching the workflow:
+
+### a. KB Folder Prompt
+
+```
+What should your engineering knowledge base folder be named?
+  Default: engineering
+  (This is where Forge writes sprint artifacts, docs, and KB files.)
+
+KB folder name [engineering]: ___
+```
+
+If the user provides a custom name (non-empty, not "engineering"), write it now:
+
+```sh
+node .forge/tools/manage-config.cjs set paths.engineering "{name}"
+```
+
+Set `kbFolder` to the chosen name (default: `"engineering"`).
+
+### b. CLAUDE.md Offer
+
+Check whether any agent instruction file exists:
+
+```sh
+ls CLAUDE.md AGENTS.md CLAUDE.local.md .cursorrules 2>/dev/null
+```
 
-**`--migrate` flag:** When `$ARGUMENTS` contains `--migrate`, run the store migration
-workflow instead of the standard init. This is the v1.0 replacement for the removed
-`/forge:migrate` command.
+If NONE of those files exists, ask:
 
-1. Check that `.forge/config.json` exists. If it does not, stop and tell the user:
-   > "Forge has not been initialised in this project. Run `/forge:init` first, then come back to `/forge:init --migrate`."
+```
+No CLAUDE.md / AGENTS.md found in this project.
+Create a minimal CLAUDE.md with Forge KB links? [Y/n]: ___
+```
 
-2. Check for `--structural` sub-flag in `$ARGUMENTS`:
-   - If present, load and run the structural migration workflow:
-     Read `"$FORGE_ROOT/meta/workflows/meta-migrate.md"` and follow it.
-   - If absent, run the standard store schema migration (Steps 1–7 of the former `migrate.md`):
-     Read `"$FORGE_ROOT/meta/workflows/meta-migrate.md"` and follow it, passing
-     `--store-schema` so it runs the schema migration path (Steps 2–7).
+Set `createClaudeMd = true` if user answers Y (or presses Enter); `false` otherwise.
+If any of those files already exists, set `createClaudeMd = false` (silently skip the offer).
 
-3. If `--dry-run` is also present, pass it through so the migration runs Steps 1–4
-   only (no writes).
+### c. Timestamp
 
-4. Do NOT proceed to the Pre-flight Plan when `--migrate` is present.
+Mint the ISO timestamp now (before workflow dispatch, in this command wrapper context):
 
-**Proceed directly to Pre-flight Plan** for all other invocations. There is no interactive mode prompt —
-the 4-phase flow is the only flow.
+```js
+isoTimestamp = new Date().toISOString()
+```
 
-### Pre-flight Plan
+## Pre-flight Plan
 
-Before executing Phase 1, emit a summary block and wait for the user to confirm
-or specify a start phase.
+Before dispatching, emit a summary and wait for confirmation:
 
 ```
 ## Forge Init — <project-name if discoverable, otherwise current directory name>
 
-4 phases will run in this session (~45 seconds non-interactive):
+4 phases will run via the wfl:init workflow:
 
   1   Collect      — 5 parallel discovery scans → config.json
-                     KB folder prompt (interactive)
+                     KB folder prompt already answered above
   2   Discover     — KB doc generation (LLM fan-out) + project-context.json
   3   Materialize  — substitute-placeholders.cjs → fully functional workflows
   4   Register     — versioning, manifest, cache, store entries, Tomoshibi
 
-Phase 1 is interactive (KB folder name prompt). Phases 2–4 are non-interactive
-and complete in under 45 seconds.
+Starting from Phase {startPhase}.
+
+Proceed? [Y/n]: ___
+```
+
+If the user declines, stop and advise them to run `/forge:init` again when ready.
+
+## Workflow Dispatch
+
+Execute the workflow:
+
+```
+workflow('wfl:init', {
+  forgeRoot,
+  kbFolder,
+  startPhase,
+  createClaudeMd,
+  isoTimestamp,
+  rawArguments: $ARGUMENTS
+})
+```
+
+Where `forgeRoot` is resolved from `.forge/config.json`'s `paths.forgeRoot` field.
 
-Start from Phase 1? [Y] or specify phase (1–4): ___
+**Workflow-tool-missing error** — if the Workflow tool is unavailable, halt immediately
+with this message:
+
+> The Workflow tool is required to run `/forge:init`. This Claude Code build does not
+> support the Workflow tool. Upgrade Claude Code and try again.
+>
+> (Alternatively, run `4ge init claude .` again to re-scaffold, then upgrade Claude Code.)
+
+Do NOT fall back to reading any other document — halt with the above message only.
+
+## Post-Workflow (on success)
+
+When `workflow('wfl:init', ...)` returns with `result.ok === true`:
+
+### Closing Banner
+
+```sh
+node .forge/tools/banners.cjs forge
+node .forge/tools/banners.cjs --subtitle "灯 SDLC ready — welcome to your Forge SDLC"
+```
+
+### Welcome Block
+
+Emit the welcome summary:
+
+```
+✓ Forge SDLC initialised
+
+Your project is now configured with:
+  · Knowledge base in {kbFolder}/
+  · Workflows, personas, and skills in .forge/
+  · Sprint-ready store in .forge/store/
+
+Next steps:
+  · /forge:new-sprint — plan your first sprint
+  · /forge:health     — verify store health
+  · /forge:plan       — plan a single task
+```
+
+### Marketplace Skills Offer
+
+If `result.skillMatches` is present and non-empty, present skill buckets by
+confidence (High → Medium → Low). For each bucket, list skill IDs and descriptions.
+Ask: `Install recommended skills? [Y/n]` (per bucket or all at once — your call).
+
+For each skill the user accepts:
+
+```sh
+node .forge/tools/manage-config.cjs set installedSkills.{skillId} true
+```
+
+Track accepted/skipped counts for the final report. **If the user skips the
+entire offer, proceed without blocking.**
+
+### KB Links Refresh
+
+Invoke the `forge:refresh-kb-links` skill via the Skill tool:
+
+```
+Skill('forge:refresh-kb-links')
 ```
 
-If the user specifies a valid phase (1–4), jump there directly.
-Any other input (including 0, 5+, or non-numeric text) re-prompts with the same table.
+If the Skill tool is unavailable or the skill fails, emit:
 
-If a `$ARGUMENTS` phase number is provided (e.g. `3`), skip the pre-flight
-table and go straight to the specified phase.
+```
+△ Could not auto-refresh KB links — run /forge:refresh-kb-links manually
+  to wire your CLAUDE.md to your new KB docs.
+```
 
-Read `$FORGE_ROOT/init/sdlc-init.md` — that document is your complete orchestration.
-Follow it exactly. It defines 4 phases:
+### Final Report
 
-1. **Collect** — 5 parallel discovery prompts, KB folder prompt, `config.json`
-2. **Discover** — KB doc fan-out + inline `project-context.json` construction
-3. **Materialize** — `substitute-placeholders.cjs` + `build-overlay.cjs` smoke test
-4. **Register** — tools, versioning, manifest, cache, store seed, Tomoshibi, `.gitignore` update (unconditional), CLAUDE.md KB-link offer
+```
+Forge init complete.
 
-The current working directory is the target project. All generated artifacts go into
-`.forge/`, the configured KB folder (default: `engineering/`), and `.claude/commands/`
-in the project.
+  KB docs generated:   {count from result.kbDocCount or "see .forge/"}
+  Workflows installed: {count from result.workflowCount or "see .forge/workflows/"}
+  Commands installed:  {count from result.commandCount or "see .claude/commands/"}
+  Skills accepted:     {acceptedCount}
+  Skills skipped:      {skippedCount}
+
+Run `/forge:health` to verify the installation.
+```
+
+## Post-Workflow (on failure)
+
+When `workflow('wfl:init', ...)` returns with `result.ok === false`, surface the
+failure as structured JSON and offer a bug report:
+
+```
+✗ Forge init failed.
+
+Failure details:
+{result.failure as formatted JSON}
+
+This looks like a Forge bug. Would you like to file a report?
+Run `/forge:report-bug` — I'll pre-fill the report from this conversation.
+```
 
 ## Arguments
 
 $ARGUMENTS
 
-## On error
+## On Error
 
 If any step above fails unexpectedly, describe what went wrong and ask:
 
-> "This looks like a Forge bug. Would you like to file a report to help improve it? Run `/forge:report-bug` — I'll pre-fill the report from this conversation."
+> "This looks like a Forge bug. Would you like to file a report to help improve it?
+> Run `/forge:report-bug` — I'll pre-fill the report from this conversation."

package/dist/forge-payload/{.base-pack/commands → commands}/new-sprint.md

@@ -3,14 +3,14 @@ name: new-sprint
 description: Elicit and structure requirements for a new sprint
 ---
 
-# /{{PREFIX}}:new-sprint
+# /forge:new-sprint
 
 Read the new-sprint workflow and follow it.
 
 ## Locate the Forge plugin
 
 ```
-FORGE_ROOT: !`echo "${CLAUDE_PLUGIN_ROOT}"`
+FORGE_ROOT: !`echo "${CLAUDE_PLUGIN_ROOT:-$(pwd)/.forge}"`
 ```
 
 ## Execute

package/dist/forge-payload/{.base-pack/commands → commands}/plan-sprint.md

@@ -3,14 +3,14 @@ name: plan-sprint
 description: Decompose sprint requirements into tasks with dependencies
 ---
 
-# /{{PREFIX}}:plan-sprint
+# /forge:plan-sprint
 
 Read the plan-sprint workflow and follow it.
 
 ## Locate the Forge plugin
 
 ```
-FORGE_ROOT: !`echo "${CLAUDE_PLUGIN_ROOT}"`
+FORGE_ROOT: !`echo "${CLAUDE_PLUGIN_ROOT:-$(pwd)/.forge}"`
 ```
 
 ## Execute

package/dist/forge-payload/{.base-pack/commands → commands}/plan.md

@@ -3,14 +3,14 @@ name: plan
 description: Design and document the implementation plan for a task
 ---
 
-# /{{PREFIX}}:plan
+# /forge:plan
 
 Read the plan workflow and follow it.
 
 ## Locate the Forge plugin
 
 ```
-FORGE_ROOT: !`echo "${CLAUDE_PLUGIN_ROOT}"`
+FORGE_ROOT: !`echo "${CLAUDE_PLUGIN_ROOT:-$(pwd)/.forge}"`
 ```
 
 ## Execute

package/dist/forge-payload/commands/rebuild.md

@@ -10,7 +10,7 @@ Re-run generation phases using the current state of the project. Use `--enrich`
 ## Locate the Forge plugin
 
 ```
-FORGE_ROOT: !`echo "${CLAUDE_PLUGIN_ROOT}"`
+FORGE_ROOT: !`echo "${CLAUDE_PLUGIN_ROOT:-$(pwd)/.forge}"`
 ```
 
 Read `.forge/config.json`. If it does not exist, stop and tell the user to run
@@ -18,7 +18,7 @@ Read `.forge/config.json`. If it does not exist, stop and tell the user to run
 
 Resolve tools from the plugin:
 ```
-FORGE_ROOT: !`echo "${CLAUDE_PLUGIN_ROOT}"`
+FORGE_ROOT: !`echo "${CLAUDE_PLUGIN_ROOT:-$(pwd)/.forge}"`
 ```
 
 All tool invocations in this command use `node "$FORGE_ROOT/tools/<tool>.cjs"`.
@@ -709,7 +709,7 @@ When `$ARGUMENTS` contains `--enrich`, run the enhancement workflow instead of r
 This is the v1.0 replacement for the removed `/forge:enhance` command.
 
 ```
-FORGE_ROOT: !`echo "${CLAUDE_PLUGIN_ROOT}"`
+FORGE_ROOT: !`echo "${CLAUDE_PLUGIN_ROOT:-$(pwd)/.forge}"`
 ```
 
 1. Check that `$FORGE_ROOT/meta/workflows/meta-enhance.md` exists. If absent:

package/dist/forge-payload/commands/remove.md

@@ -21,7 +21,7 @@ nothing is deleted until you confirm explicitly.
 Read the configured KB path and project prefix from config:
 
 ```sh
-FORGE_ROOT: !`echo "${CLAUDE_PLUGIN_ROOT}"`
+FORGE_ROOT: !`echo "${CLAUDE_PLUGIN_ROOT:-$(pwd)/.forge}"`
 KB_PATH: !`node -e "try{console.log(require('./.forge/config.json').paths.engineering)}catch{console.log('engineering')}"`
 PREFIX: !`node -e "try{console.log(require('./.forge/config.json').project.prefix.toLowerCase())}catch{console.log('')}"`
 ```

package/dist/forge-payload/commands/repair.md

@@ -19,7 +19,7 @@ $ARGUMENTS
 
 1. Resolve the plugin root:
    ```
-   FORGE_ROOT: !`echo "${CLAUDE_PLUGIN_ROOT}"`
+   FORGE_ROOT: !`echo "${CLAUDE_PLUGIN_ROOT:-$(pwd)/.forge}"`
    ```
 
 2. Run the four-phase repair workflow below. If `--dry-run` is in `$ARGUMENTS`,

package/dist/forge-payload/commands/report-bug.md

@@ -30,7 +30,7 @@ gh is not installed or not authenticated.
 Collect the following in parallel:
 
 ```
-forge_version:  !`cat "${CLAUDE_PLUGIN_ROOT}/.claude-plugin/plugin.json" 2>/dev/null | grep '"version"' | head -1 | sed 's/.*: *"\(.*\)".*/\1/'`
+forge_version:  !`cat "${CLAUDE_PLUGIN_ROOT:-$(pwd)/.forge}/.claude-plugin/plugin.json" 2>/dev/null | grep '"version"' | head -1 | sed 's/.*: *"\(.*\)".*/\1/'`
 node_version:   !`node --version 2>/dev/null || echo "N/A"`
 os_info:        !`uname -srm 2>/dev/null || echo "N/A"`
 forge_config:   !`cat ".forge/config.json" 2>/dev/null | head -30 || echo "No .forge/config.json found"`

package/dist/forge-payload/commands/reset.md

@@ -0,0 +1,117 @@
+---
+name: reset
+description: Rewind a halted/blocked task, bug, or sprint to an earlier pipeline phase — a guardrailed, integrity-checked state reset. Use when a pipeline halted and you need to resume from a chosen phase, or to reopen committed work for rework.
+allowed-tools:
+  - Bash
+---
+
+# /forge:reset
+
+Guardrailed, natural-language **pipeline-state reset**. Rewinds a task, bug, or
+sprint to an earlier phase so its pipeline can be resumed — after computing and
+**confirming the cross-entity referential-integrity implications**. This never
+mutates the store until you explicitly confirm the proposed plan.
+
+> Deterministic counterpart: `4ge reset <id> --to <phase>` (forge-cli) does a
+> single-entity task/bug reset without the NLP/integrity layer. This command
+> adds intent parsing, sprint cascades, and the integrity gate.
+
+## Arguments
+
+`$ARGUMENTS` is a free-text request, e.g.:
+- "put FORGE-S32-T03 back to implement"
+- "redo the fix for FORGE-BUG-046 from triage"
+- "reopen sprint FORGE-S32 from task T03"
+
+## Resolve plugin root
+
+```
+FORGE_ROOT: !`node -e "console.log(require('./.forge/config.json').paths.forgeRoot)"`
+```
+
+## Step 1 — Interpret the request
+
+From `$ARGUMENTS`, determine:
+- **entity kind**: `task` (`*-S##-T##`), `bug` (`*-BUG-###`), or `sprint`
+  (`*-S##`). A `-BUG-` segment ⇒ bug.
+- **id**: the canonical record id. If the id is partial or ambiguous, resolve it
+  first with `node "$FORGE_ROOT/tools/store-cli.cjs" query <text>` (or
+  `store-query-nlp`) and pick the unambiguous match; if ambiguous, ask the user.
+- **target phase** (`--to`):
+  - task: `plan | review-plan | implement | review-code | validate | approve | writeback | commit`
+  - bug: `triage | plan-fix | review-plan | implement | review-code | approve | commit`
+  - sprint: requires a **from-task** (the member task to rewind from); the
+    cascade is computed for you.
+
+If the target phase is missing or not valid for the entity kind, ask the user
+rather than guessing.
+
+## Step 2 — Compute the plan + integrity implications (READ-ONLY)
+
+```sh
+# task / bug
+node "$FORGE_ROOT/tools/reset-plan.cjs" --entity <kind> --id <id> --to <phase> --json
+# sprint
+node "$FORGE_ROOT/tools/reset-plan.cjs" --entity sprint --id <sprintId> --from-task <taskId> --json
+```
+
+`reset-plan.cjs` is a **pure planner** — it reads the store and computes the
+transitions + warnings, and writes nothing. Parse its JSON:
+
+- `entity` — kind, id, currentStatus, targetStatus/targetPhase.
+- `rewind` / `forceRequired` — whether this is a backward transition (needs
+  `--force`).
+- `transitions[]` — the exact `{id, kind, from, to}` status changes a confirmed
+  reset would apply (for a sprint: the sprint reopen + each cascade task).
+- `cascade[]` — (sprint) the member task ids that rewind.
+- `warnings[]` — referential-integrity implications, each with a `code`:
+  - `committed-work` — the entity is committed/fixed; **its git commit is NOT
+    reverted** by the reset. The working tree may diverge.
+  - `sprint-incoherent` / `sprint-reopen` — the parent sprint is completed; it
+    must reopen to `active` (sprint resets do this automatically).
+  - `dependents-affected` — other tasks depend on this id and are past planning;
+    they now rest on rewound work and may need re-validation. (`ids` lists them.)
+
+If `reset-plan.cjs` exits non-zero, surface its `error` and stop.
+
+## Step 3 — Present the plan and REQUIRE confirmation
+
+Show the user, in plain language:
+1. The transition(s) that will be applied (`from → to` for each id).
+2. Every warning, especially `committed-work` and `dependents-affected` — these
+   are the referential-integrity risks they are accepting.
+3. The note that this does **not** revert git commits or re-run anything; it
+   only rewinds store status + (for `4ge` users) the resume-state cache.
+
+**Do not proceed without an explicit "yes".** This is a destructive,
+hard-to-reverse store mutation. If the user hesitates, stop.
+
+## Step 4 — Apply (only after confirmation)
+
+Apply each transition in `transitions[]` in order, via store-cli. Backward
+transitions and terminal states (`blocked`, `committed`, `fixed`) require the
+operator-gated force flag:
+
+```sh
+FORGE_ALLOW_FORCE=1 node "$FORGE_ROOT/tools/store-cli.cjs" \
+  update-status <kind> <id> status <to> --force
+```
+
+For a **sprint** reset, apply the sprint reopen first, then each cascade task.
+
+After the status transitions land:
+- Tell the user the reset is complete and how to resume:
+  - task → `/forge:run-task <id>` (or `/forge:implement <id>`)
+  - bug → `/forge:fix-bug <id>`
+  - sprint → `/forge:run-sprint <id>`
+- If they drive Forge through **forge-cli** (`4ge`), the resume-state cache
+  (`.forge/cache/*-state.json`) is rewound by `4ge reset <id> --to <phase>` —
+  the LLM route here only owns the store side. Mention this so a `4ge` user runs
+  `4ge reset` (or deletes the stale cache) before resuming.
+
+## Iron rule
+
+Never apply a transition `reset-plan.cjs` did not propose, and never skip the
+confirmation gate. The planner is the single source of truth for what a reset
+touches; this command's only job is to explain it and, on a clear yes, apply
+exactly that.

package/dist/forge-payload/{.base-pack/commands → commands}/retro.md

@@ -3,14 +3,14 @@ name: retro
 description: Produce sprint retrospective and lessons-learned document
 ---
 
-# /{{PREFIX}}:retro
+# /forge:retro
 
 Read the retro workflow and follow it.
 
 ## Locate the Forge plugin
 
 ```
-FORGE_ROOT: !`echo "${CLAUDE_PLUGIN_ROOT}"`
+FORGE_ROOT: !`echo "${CLAUDE_PLUGIN_ROOT:-$(pwd)/.forge}"`
 ```
 
 ## Execute

package/dist/forge-payload/{.base-pack/commands → commands}/review-code.md

@@ -3,14 +3,14 @@ name: review-code
 description: Code quality review of a completed task implementation
 ---
 
-# /{{PREFIX}}:review-code
+# /forge:review-code
 
 Read the review-code workflow and follow it.
 
 ## Locate the Forge plugin
 
 ```
-FORGE_ROOT: !`echo "${CLAUDE_PLUGIN_ROOT}"`
+FORGE_ROOT: !`echo "${CLAUDE_PLUGIN_ROOT:-$(pwd)/.forge}"`
 ```
 
 ## Execute

package/dist/forge-payload/{.base-pack/commands → commands}/review-plan.md

@@ -3,14 +3,14 @@ name: review-plan
 description: Architectural review of a task implementation plan
 ---
 
-# /{{PREFIX}}:review-plan
+# /forge:review-plan
 
 Read the review-plan workflow and follow it.
 
 ## Locate the Forge plugin
 
 ```
-FORGE_ROOT: !`echo "${CLAUDE_PLUGIN_ROOT}"`
+FORGE_ROOT: !`echo "${CLAUDE_PLUGIN_ROOT:-$(pwd)/.forge}"`
 ```
 
 ## Execute

package/dist/forge-payload/{.base-pack/commands → commands}/run-sprint.md

@@ -3,14 +3,14 @@ name: run-sprint
 description: Execute all tasks in a sprint (sequential or parallel)
 ---
 
-# /{{PREFIX}}:run-sprint
+# /forge:run-sprint
 
 Read the run-sprint workflow and follow it.
 
 ## Locate the Forge plugin
 
 ```
-FORGE_ROOT: !`echo "${CLAUDE_PLUGIN_ROOT}"`
+FORGE_ROOT: !`echo "${CLAUDE_PLUGIN_ROOT:-$(pwd)/.forge}"`
 ```
 
 ## Execute

package/dist/forge-payload/{.base-pack/commands → commands}/run-task.md

@@ -3,14 +3,14 @@ name: run-task
 description: Orchestrate the full plan→implement→review→approve pipeline for a task
 ---
 
-# /{{PREFIX}}:run-task
+# /forge:run-task
 
 Read the run-task workflow and follow it.
 
 ## Locate the Forge plugin
 
 ```
-FORGE_ROOT: !`echo "${CLAUDE_PLUGIN_ROOT}"`
+FORGE_ROOT: !`echo "${CLAUDE_PLUGIN_ROOT:-$(pwd)/.forge}"`
 ```
 
 ## Execute

package/dist/forge-payload/commands/status.md

@@ -12,7 +12,7 @@ Quick overview of the current sprint and task state.
 ## Locate plugin root
 
 ```
-FORGE_ROOT: !`echo "${CLAUDE_PLUGIN_ROOT}"`
+FORGE_ROOT: !`echo "${CLAUDE_PLUGIN_ROOT:-$(pwd)/.forge}"`
 ```
 
 ## Locate project config