implement-status
Skill3 starsChanged 6 months agoMITView on GitHub
Show progress for the currently-running `/build-tdds` job. Prints a one-shot snapshot (current TDD, stage, an estimate-labeled percent, per-TDD statuses, log/PR pointers) and, for a live watch, hands you a one-line `!bash …/scripts/status.sh --follow` command to paste yourself (Ctrl-C exits the watch; the build is unaffected). Read-only — no pause/resume/cancel. Invoke with /implement-status.
In this repository6 other files it gives its agents.
- adr-newSkill
- bootstrap-projectSkill
- build-tddsSkill
- prd-authorSkill
- tdd-authorSkill
- ux-authorSkill
The file
We split this skill into the rules it gives an agent
4 rules. About 0 are common, 0 are distinctive, and 4 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈0
Distinctive · estimate
This author's own choices. Worth a look.
4
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 31
If no /build-tdds run is active, the skill says so plainly and (if a previous run exists) summarizes the last run's final state.
- Very specific: line 37
When the most recent run is halted (state paused, blocked, or failed), the snapshot switches to a one-screen halt context (FR-64): the halt …
- Very specific: line 66
Add --logdir <dir> to inspect a specific historical run dir instead of the active one.
- Very specific: line 78
DO NOT call status.sh --follow through the Bash tool from this skill — it is an endless foreground loop that would block the session.
≈ “Common” is an estimate: at least 1 in 5 of the 372,429 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Implement status
A read-only progress view for the /build-tdds run. The runner
writes a structured run-state record under
docs/tdd/.implement-logs/<ts>/state.d/ (FR-27); this skill renders it.
What you see
The snapshot derives the rendered view fresh on each call from the per-TDD fragments — there is no second artifact that can drift from the record. It shows:
- Header — mode (sequential / combined / parallel), integration branch, elapsed time, run state.
- Rollup —
<completed> done / <total> · ~<P>% (estimate)— the percent is always suffixed(estimate)and never reads 100% while any TDD is non-terminal (FR-30 honesty). Failed / blocked / skipped counts are reported separately when non-zero. - Per-TDD table — queue position, slug, status
(
pending|building|verifying|reviewing|done|failed|blocked|skipped), current stage (build|test-first|verify|verify-runtime|review|flip), and the TDD's branch / PR URL when known. - Current focus — the in-progress TDD and its stage, when a TDD is non-terminal.
If no /build-tdds run is active, the skill says so plainly and (if a previous
run exists) summarizes the last run's final state. It never reports false
progress.
Halted runs
When the most recent run is halted (state paused, blocked, or failed), the
snapshot switches to a one-screen halt context (FR-64): the halt cause from the
closed enum, the triggering finding reference, and the deterministic next-action
options, each on its own numbered line. Halted-run rendering fits 24×80 by
default; to see full logs use cat docs/tdd/.implement-logs/<runid>/REPORT. The
Resume: /build-tdds --resume <runid> line appears only when the cause is a
recoverable (paused-state) one.
A run whose state is interrupted (TDD 0030 §3) did not exit cleanly — the
runner died mid-gate, leaving one or more TDDs orphaned in a non-terminal status.
The snapshot renders a distinct "the run did not exit cleanly" banner naming each
orphaned TDD + gate and pointing at /build-tdds --resume <runid>; it is never
reported as done.
--follow watch mode in a non-interactive background job: use kill -TERM (or
-HUP/-QUIT), not kill -INT — SIGINT is silently un-trappable in that launch
mode per POSIX. SIGINT still works correctly in the foreground.
Run it
For the on-demand snapshot, run the renderer once via a single Bash call and relay the output:
_tl_src="${CLAUDE_PLUGIN_ROOT:-${GROK_PLUGIN_ROOT:-}}"
. "${_tl_src}/scripts/lib/plugin-root.sh"
bash "$(tl_plugin_root)/scripts/status.sh"Add --logdir <dir> to inspect a specific historical run dir instead of the
active one.
Live watch (do NOT invoke from this skill)
For a live watch that refreshes until you press Ctrl-C, hand the user the
following line for them to paste themselves as a foreground ! command:
!bash -c '_tl_src="${CLAUDE_PLUGIN_ROOT:-${GROK_PLUGIN_ROOT:-}}"; . "${_tl_src}/scripts/lib/plugin-root.sh" && bash "$(tl_plugin_root)/scripts/status.sh" --follow'DO NOT call status.sh --follow through the Bash tool from this skill — it is
an endless foreground loop that would block the session. The user runs it in
their TUI directly; Ctrl-C in their terminal exits the watch cleanly. The watch
is read-only — it never signals the build PID, never edits files, and offers
no pause / resume / cancel.
Where the record lives
docs/tdd/.implement-logs/<ts>/state.d/run.json— run-level rollup + identity.docs/tdd/.implement-logs/<ts>/state.d/<slug>.json— one fragment per queued TDD, updated atomically at every status / stage transition.docs/tdd/.implement-logs/latest— a symlink pointing at the active run's<ts>directory;status.shresolves it by default.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

