From a3da3d7ace8ef62670f43fa807d91ca24e0306db Mon Sep 17 00:00:00 2001 From: "github-actions[bot]" <41898282+github-actions[bot]@users.noreply.github.com> Date: Mon, 11 May 2026 11:43:21 +0000 Subject: [PATCH] docs: fix agent-team documentation after dispatch-input hardening MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit - README.md: fix incorrect 7th label in uninstall section (`state:in-progress` → `agent-team:reviewed`); expand to explicit per-label `gh label delete` commands for clarity - catalog/agent-team/README.md: expand "Input propagation" gotcha to document the fail-loud behavior added in #69 — unresolved template literal detection, issue_number-missing vs. present handling, and the pr_number blank/literal treatment for first-dispatch vs. kickback - CONTRIBUTING.md: fix "seven dogfooded workflows" → six (three agentic + three CI), and clarify that only the agentic workflows are gh-aw managed Co-Authored-By: Claude Sonnet 4.6 --- CONTRIBUTING.md | 2 +- README.md | 11 ++++++++++- catalog/agent-team/README.md | 5 ++++- 3 files changed, 15 insertions(+), 3 deletions(-) diff --git a/CONTRIBUTING.md b/CONTRIBUTING.md index 1c2721e..5ecc1cb 100644 --- a/CONTRIBUTING.md +++ b/CONTRIBUTING.md @@ -85,7 +85,7 @@ Never test by committing untested changes to `main`. The installed workflows run ## Workflow files -The `.github/workflows/` directory contains seven dogfooded workflows. These are managed by `gh aw` — do not edit `.lock.yml` files by hand except to apply the OAuth tweak described in [skills/install-workflow/auth.md](skills/install-workflow/auth.md). +The `.github/workflows/` directory contains six workflows: three dogfooded agentic workflows (`daily-repo-status`, `update-docs`, `weekly-research`) and three standard CI workflows (`agentics-maintenance`, `ci-tests`, `copilot-setup-steps`). Only the agentic workflows are managed by `gh aw` — do not edit their `.lock.yml` files by hand except to apply the OAuth tweak described in [skills/install-workflow/auth.md](skills/install-workflow/auth.md). If a workflow `.md` source needs changing: diff --git a/README.md b/README.md index e18e51a..a675b32 100644 --- a/README.md +++ b/README.md @@ -80,7 +80,16 @@ To remove workflows this plugin installed into your target repo: - `gh aw remove ` for each installed workflow (deletes both the `.md` source and the compiled `.lock.yml`), then commit the deletion. - `gh secret delete CLAUDE_CODE_OAUTH_TOKEN` — or `ANTHROPIC_API_KEY`, whichever path you used — to unset the auth secret. -- For `agent-team` specifically, also delete the seven labels: `gh label delete agent-team` plus `gh label delete state:` for each of `plan-needed`, `impl-needed`, `review-needed`, `done`, `blocked`, and `in-progress`. +- For `agent-team` specifically, also delete the seven labels: + ```bash + gh label delete agent-team + gh label delete state:plan-needed + gh label delete state:impl-needed + gh label delete state:review-needed + gh label delete state:done + gh label delete state:blocked + gh label delete agent-team:reviewed + ``` Nothing else is persisted — the plugin writes only to your target repo (under user approval) and holds no local state outside Claude Code's own plugin directory. diff --git a/catalog/agent-team/README.md b/catalog/agent-team/README.md index 5490837..03215bc 100644 --- a/catalog/agent-team/README.md +++ b/catalog/agent-team/README.md @@ -111,7 +111,10 @@ Then apply the OAuth token tweak to each `.lock.yml` per [`skills/install-workfl - **Concurrency**: each workflow uses `concurrency: group: agent-team-issue-${issue_number}` so only one role runs at a time per issue. - **Max iterations**: default 3 (reviewer kickback → implementer). The counter lives on the `iteration` input passed through the dispatch chain, bumped exclusively by the reviewer on kickback. -- **Input propagation**: planner / implementer / reviewer must fail loudly if required `workflow_dispatch` inputs are missing. Do not rely on label search or recent-activity inference as a fallback. +- **Input propagation**: planner / implementer / reviewer fail loudly if required `workflow_dispatch` inputs are missing **or** still appear as unresolved template literals (e.g. the raw string `${{ github.event.inputs.issue_number }}`). Never infer a missing value from labels, recent activity, or search results. Exact behavior: + - `issue_number` present but empty/unresolved → add `state:blocked`, post `🛑 agent-team: workflow_dispatch inputs were not propagated. Re-dispatch with valid inputs.` + - `issue_number` itself missing → call `missing_data` / `report_incomplete` to fail loudly. + - `pr_number` (implementer only) blank or still a literal → treated as "no existing PR"; a new branch + PR is created. Both are valid first-dispatch states. - **Non-UI only**: no screenshot capture. Reviewer validates via tests/CI status + reading the diff. - **Cost**: a single task can easily spend 4× the tokens of a monolithic workflow. Set `timeout-minutes` conservatively and monitor the first few runs. - **No auto-merge**: the reviewer approves but never merges. Humans merge.