sequant 2.10.0 → 2.12.0

package/templates/skills/spec/references/recommended-workflow.md

@@ -1,6 +1,17 @@
 # Recommended Workflow Format
 
-This document shows the expected output format for the `## Recommended Workflow` section in `/spec` output. The `parseRecommendedWorkflow()` function parses this format to determine which phases to execute.
+This document shows the expected output format for the `## Recommended Workflow` section in `/spec` output.
+
+## Resolution chain (#921)
+
+`sequant run` resolves phases through an ordered chain, not `parseRecommendedWorkflow()` alone:
+
+1. **`SEQUANT_SPEC` marker** — a structured HTML comment in the posted plan comment, e.g. `<!-- SEQUANT_SPEC: {"phases":["testgen","exec","qa"],"qualityLoop":true} -->`. This is the primary, durable channel — always emit it alongside the prose section below.
+2. **Comment prose** — `parseRecommendedWorkflow()` applied to the plan comment body (same format as this doc).
+3. **Chat text** — the same parser applied to the spec agent's chat output. Nondeterministic: only present if the agent happens to restate the section in chat rather than posting via a body file (#814).
+4. **Label fallback** — `detectPhasesFromLabels()`. Can never produce `testgen` or `security-review`.
+
+The marker's `phases` array excludes `spec` (it already ran) and must name only registered phases — an unknown phase name invalidates the whole marker and falls through to step 2.
 
 ## Format
 
@@ -10,6 +21,8 @@ This document shows the expected output format for the `## Recommended Workflow`
 **Phases:** spec → exec → qa
 **Quality Loop:** disabled
 **Reasoning:** Brief explanation of why this workflow was chosen.
+
+<!-- SEQUANT_SPEC: {"phases":["exec","qa"],"qualityLoop":false} -->
 ```
 
 ## Examples

package/templates/skills/test/SKILL.md

@@ -582,7 +582,7 @@ Create structured test results:
 ### 3.2 GitHub Comment
 
 **If orchestrated (SEQUANT_ORCHESTRATOR is set):**
-- Skip posting GitHub comment (orchestrator handles summary)
+- Skip posting this skill's own GitHub comment — no per-phase comment is posted under `sequant run`; test results surface through the run summary and the PR body (#964)
 - Include test summary in output for orchestrator to capture
 - Let orchestrator aggregate results across phases
 

package/templates/skills/testgen/SKILL.md

@@ -6,6 +6,7 @@ metadata:
   author: sequant
   version: "1.0"
 allowed-tools:
+  - Bash(npx sequant worktree:*)
   - Read
   - Write
   - Edit
@@ -530,17 +531,33 @@ If an AC has verification method "N/A - Trivial", skip test generation and note
 
 ### Step 4: Locate Feature Worktree
 
-If generating file-based tests (Unit Test, Integration Test), find the worktree:
+If generating file-based tests (Unit Test, Integration Test), find the worktree.
 
-```bash
-git worktree list | grep -E "feature.*<issue-number>" || true
-```
+<!-- BEGIN: worktree-standalone-lookup (#899) -->
+
+Resolve it through git, not the filesystem:
 
-Or check:
 ```bash
-ls ../worktrees/feature/<issue-number>-*/
+WORKTREE="$(npx sequant worktree resolve <issue-number>)" || {
+  echo "❌ HALT: no worktree for #<issue-number> in this repository."
+  exit 1
+}
+cd "$WORKTREE"
 ```
 
+`sequant worktree resolve` reads `git worktree list` in the current repository
+— which reports only *this* repo's worktrees — and selects on the **branch**
+git reports, not the directory name.
+
+**Do not glob `../worktrees/feature/<issue-number>-*`, and do not grep
+`git worktree list` for the issue number.** The first matches across sibling
+repositories, which share that directory; the second matches the printed path,
+so it keys on the directory slug — and a slug can drift from its own branch
+after a rename. Because this skill **writes test files**, landing in the wrong
+tree scatters stubs into an unrelated project.
+
+<!-- END: worktree-standalone-lookup (#899) -->
+
 Create test directories if needed:
 ```bash
 mkdir -p __tests__/integration

package/templates/agents/sequant-explorer.md

@@ -1,24 +0,0 @@
----
-name: sequant-explorer
-description: Codebase exploration agent for sequant /spec phase. Searches for existing patterns, components, database schemas, and file structures. Use when gathering context before planning a feature implementation.
-# Note: per anthropics/claude-code#43869 this is currently a no-op; agent runs on parent's model
-model: haiku
-maxTurns: 15
-tools:
-  - Read
-  - Grep
-  - Glob
----
-
-You are an exploration agent for the sequant development workflow.
-
-Your job is to search the codebase for existing patterns, components, and structures relevant to a planned feature.
-
-Rules:
-- Search thoroughly across relevant directories
-- Report findings in structured format: file paths, patterns discovered, recommendations
-- Do NOT modify any files
-- Do NOT run shell commands
-- Send results back via SendMessage when complete
-- All grep commands must use `|| true` to prevent exit code 1 on zero matches
-- Focus on actionable findings that inform implementation decisions