Beta skills now use disable-model-invocation: true to prevent accidental
auto-triggering. Descriptions written as future stable descriptions with
[BETA] prefix for clean promotion. Updated solutions doc and AGENTS.md
promotion checklist to include removing the field.
- Beta plans use -beta-plan.md suffix to avoid clobbering stable plans
- Fix internal references in beta skills to use beta names consistently
- Add beta skills section to AGENTS.md with promotion checklist
Create separate beta skills instead of gating existing ones. Stable
ce:plan and deepen-plan are restored to main versions. Beta skills
reference each other and work standalone outside lfg/slfg orchestration.
- Surface deferred implementation questions and scope boundaries
- Use per-unit Patterns and Verification fields for task execution
- Add execution strategy: inline, serial subagents, or parallel
- Reframe Swarm Mode as Agent Teams with opt-in requirement
- Make tool references platform-agnostic
- Remove plan checkbox editing during execution
Restructures ce:plan around a decisions-first philosophy:
- Replace issue-template output with durable implementation plans
- Add blocker classification gate for upstream requirements (R11-R13)
- Replace MINIMAL/MORE/A LOT with Lightweight/Standard/Deep
- Add planning bootstrap fallback with ce:brainstorm recommendation
- Remove all implementation code, shell commands, and executor litter
- Make SpecFlow conditional for Standard/Deep plans
- Keep research agents, brainstorm-origin integration, and handoff options
- Restore origin doc completeness checks, user signal gathering,
research decision examples, filename examples, stakeholder awareness,
and mermaid diagram nudges from the old skill
plugin.json and marketplace.json were stuck at 2.40.0 while root
package.json was already at 2.41.0. Skill count was listed as 47
but actual count is 42. README still had stale "Commands | 23"
row from before the commands→skills migration in v2.39.0.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
New issue-intelligence-analyst agent that fetches GitHub issues via
gh CLI, clusters by root-cause themes, and returns structured analysis
with trend direction, confidence scores, and source mix. Designed for
both ce:ideate integration and standalone use.
Agent design:
- Priority-aware fetching with label scanning for focus targeting
- Truncated bodies (500 chars) in initial fetch to avoid N+1 calls
- Single gh call per fetch, no pipes or scripts (avoids permission spam)
- Built-in --jq for all field extraction and filtering
- Mandatory structured output with self-check checklist
- Accurate counts from actual data, not assumptions
- Closed issues as recurrence signal only, not standalone evidence
ce:ideate gains:
- Issue-tracker intent detection in Phase 0.2
- Conditional agent dispatch in Phase 1 (parallel with existing scans)
- Dynamic frame derivation from issue clusters in Phase 2
- Hybrid strategy: cluster-derived frames + default padding when < 4
- Resume awareness distinguishing issue vs non-issue ideation
- Numbered table format for rejection summary in ideation artifacts
Reduce per-agent idea target from 10 to 7-8 based on real usage data
showing ideas 8-11 were speculative tail that rarely survived filtering.
This keeps the unique candidate pool manageable (~20-30 after dedup)
while preserving frame diversity across 4-6 agents. Also add scannable
overview line before detail blocks in Phase 4, and clarify foreground
dispatch and native tool usage in Phase 1.
- Clarify sub-agent volume: each agent targets ~10 ideas (40-60 raw, ~30-50 after dedupe)
- Reframe ideation lenses as starting biases, not constraints, to encourage cross-cutting ideas
- Add orchestrator synthesis step between merge/dedupe and critique
- Improve skill description with specific trigger phrases for better auto-discovery
- Update argument-hint to be user-facing ("feature, focus area, or constraint")
- Position ideate as optional entry point in workflow diagram, not part of core loop
- Update plugin metadata and README with new skill counts and descriptions
Research agents (repo-research-analyst, git-history-analyzer,
best-practices-researcher, framework-docs-researcher) were using
shell commands like find, rg, cat, and chained pipelines for routine
codebase exploration. This triggers permission prompts in Claude Code
and degrades the user experience when these agents run as sub-agents.
Updated all research agents with platform-agnostic tool selection
guidance that prefers native file-search/glob, content-search/grep,
and file-read tools over shell equivalents. Shell is now reserved for
commands with no native equivalent (ast-grep, bundle show, git).
Git-history-analyzer additionally limits shell to one simple git
command per call with no chaining or piping.
Added tool selection rules to AGENTS.md so future agents follow
the same pattern by default.
* docs: capture codex skill prompt model
* fix: align codex workflow conversion
* chore: remove deprecated workflows:* skill aliases
The workflows:brainstorm, workflows:plan, workflows:work, workflows:review,
and workflows:compound aliases have been deprecated long enough. Remove them
and update skill counts (46 → 41) across plugin.json, marketplace.json,
README, and CLAUDE.md.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
---------
Co-authored-by: Trevin Chow <trevin@trevinchow.com>
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* feat: make PR/commit attribution specific to model, harness, and plugin version
Replace generic "Generated with Claude Code" footer with dynamic attribution
that includes the actual model name, harness tool, and plugin version. LLMs
fill in their own values at commit/PR time. Subagents are explicitly
instructed to do the same.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* style: format attribution substitution guide as table
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* style: rename badge to "Compound Engineering v[VERSION]"
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* feat: add context window and thinking level to attribution
Separate MODEL into MODEL, CONTEXT, and THINKING placeholders
so each detail is its own table row and easier to read.
Co-Authored-By: Claude Opus 4.6 (1M context, extended thinking) <noreply@anthropic.com>
* style: badge on its own line, model details on next line in PR template
Co-Authored-By: Claude Opus 4.6 (1M context, extended thinking) <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
The Phase 0 vibes-based heuristic warned users and forced a mode choice
even with 80%+ context remaining. Full mode now runs by default;
compact-safe mode stays available on explicit request.
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
The model was offering "commit to current branch" on main instead
of "create a branch and PR." Added explicit branch detection step
and "Do NOT commit directly to main" instruction.
Let the agent generate a reasonable branch name based on context
and repo conventions instead of prescribing a date-based format
that would collide on multiple runs per day.
Handles committing changes at the end of a refresh run so doc
maintenance doesn't sit uncommitted. Detects git context and adapts:
autonomous mode uses sensible defaults (branch + PR on main, separate
commit on feature branches), interactive mode presents options. Always
selectively stages only compound-refresh files to avoid mixing with
in-progress feature work.
The file-tools-over-bash instruction was in the orchestrator's
context but not passed to spawned subagents. Changed to an explicit
quoted instruction block that must be included in each subagent's
task prompt so it's visible to the subagent, not just the orchestrator.
Auto-archive now requires both the implementation AND the problem
domain to be gone. If referenced files are deleted but the application
still deals with the same problem (auth, payments, migrations), the
learning should be Replace'd not Archive'd — the knowledge gap needs
to be filled. Uses agent reasoning about concepts, not mechanical
keyword searches.
The model was generating findings internally then outputting a
one-line summary. Added explicit instructions that the full report
must be printed as text output — every file, every classification,
every action. In autonomous mode, the report is the sole deliverable
and must be self-contained and complete.
- Restructure Phase 3 with explicit autonomous skip section that says
"do not ask, do not present, do not wait" before any interactive
instructions
- Add autonomous caveats to Core Rules 4, 7, 8 which previously had
unconditional "ask the user" language
- Clarify that missing referenced files is unambiguous Archive evidence,
not a doubt case requiring user input
Instead of requiring write permissions, autonomous mode attempts
writes and gracefully falls back to recommendations when denied.
Report splits into Applied (succeeded) and Recommended (could not
write) sections. Read-only invocations produce a maintenance plan.
Support mode:autonomous argument for unattended/scheduled runs.
In autonomous mode: skip all user questions, apply safe actions
directly, mark ambiguous cases as stale with conservative confidence,
and generate a detailed report for after-the-fact human review.
- Broad scope triage: inventory + impact clustering + spot-check drift
for 9+ docs, recommends highest-impact area instead of blind ask
- Drift classification: sharp boundary between Update (fix references
in-skill) and Replace (subagent writes successor learning)
- Replacement subagents: sequential subagents write new learnings using
ce:compound's document format with investigation evidence already
gathered, avoiding redundant research
- Stale fallback: when evidence is insufficient for a confident
replacement, mark as stale and recommend ce:compound later
