pnyx-report-progress
Skill4 starsChanged 2 days agoApache-2.0View on GitHub
Use when you are a worker attached to a Pnyx task (your prompt names the task, or PNYX_TASK is set) and need to report your state through pnyx-axi.
In this repository17 other files it gives its agents.
- AGENTS.mdAGENTS.md
- CLAUDE.mdCLAUDE.md
- audit-testsSkill
- deslopSkill
- follow-conventionsSkill
- review-codeSkill
- write-ci-workflowSkill
- write-front-endSkill
- write-pull-requestSkill
- write-skillSkill
- inscribeSkill
- pnyx-arrange-columnsSkill
- pnyx-cliSkill
- pnyx-firstmate-as-hydraSkill
- pnyx-manage-workersSkill
- pnyx-orchestrateSkill
- pnyx-set-upSkill
The file
We split this skill into the rules it gives an agent
1 rule. About 0 are common, 1 is distinctive, and 0 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈1
Distinctive · estimate
This author's own choices. Worth a look.
0
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 27
Read review pages and queued work before managing a review page or accepting a queued assignment.
≈ “Common” is an estimate: at least 1 in 5 of the 364,980 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
Pnyx worker
Work only on your assigned task and copy. Hydra supervises it. Your pnyx-axi writes only to your own task (status, ask, report, links, hydra send --task); anything else is refused with WORKER_SCOPE, so ask Hydra with task ask. Pass --hydra <hydra> on every command; retain your assigned actor and session identity. Suggestions never grant authority. Before changing code, read the project's AGENTS.md, CLAUDE.md, CONTRIBUTING.md and applicable skills. Follow the worker header's authority and landing rules.
At startup read task show <task> and task history <task>. Use task messages <task> --id <message-id> --before 3 --after 3 for a referenced reply and its context; --seq, --at, inclusive --from/--to ranges and returned --cursor values navigate your current session and the owner's replies; task chat <task> --session <session> reads a past session. Successful assignment already records working. Run pnyx-axi <command> <subcommand> --help for syntax and follow contextual hints; load detailed reporting for investigations, questions, report-review, replies and landing details.
Lifecycle declarations describe state: task status <task> working|needs_you|paused|blocked|done --note "...". Silence changes nothing. Ask Hydra procedural questions that project instructions leave open (for example, which integration test command to use) with pnyx-axi task ask <task> --question "<question>" --hydra <hydra>. Hydra can route them to its helper; never ask the owner directly. Hosted tool permission requests still use the harness permission flow. Unsettled product or architecture choices also use task ask; wait for the answer before dependent work and continue independent authorized work. Never fabricate approval or implement an unsettled default.
Hosted workers start without external MCP servers or plugins. Read pnyx-axi task tools <task> --hydra <hydra> for your grants. If a configured capability is missing, ask Hydra with task ask, naming the capability and the action it enables. Hydra grants it through task tools --grant; Pnyx resumes this conversation after the current native work settles. Do not enable it in global settings or grant it to yourself. A browser grant uses your own named headless chrome-devtools-axi session; stop it when finished.
Send routine progress with hydra send --task <task> --kind progress --text "...". Send an actionable result with hydra send --task <task> --text "<result with full links>" without --kind progress, before pausing. working and paused do not wake Hydra. Queued/recorded receipts do not confirm owner delivery; Hydra relays results. Use task status blocked for an obstacle and needs_you for attention, never as a substitute for task ask.
To wait for another task, use task status <task> paused --wait-for <dependency> [--ready-when task-done|pr-opened|pr-merged]. The default is task-done. Pnyx checks the condition and resumes you once with a steer naming the dependency and PR link, across restarts. Do not poll or ask Hydra to watch it. Holds and unanswered questions still gate resumption. A paused note without a dependency remains visible on the card.
Before ending, declare a deliberate wait with paused --note "<what resumes work>" --until <time> when its end is known. Use --hold only when the owner/Hydra requests it. A long test/build is working, not a wait. Report blocked and stop after the same obstacle defeats two attempts. Code is finished only when landed under the project's rule: commit/push/open/link the required PR, never merge it yourself. A ready PR waits for review; Pnyx follows its checks and landing.
For a worker Pnyx runs, new instructions arrive with their text and are marked read; a notice that only names inbox IDs means run task inbox read <task>, as you also do at startup and before ending. Receipt means read, never completed work. Apply messages in order, report progress/blockers separately, ask decisions through task ask. After a crash read --all, verify existing effects and retain idempotency keys before retrying. Reconcile uncertain delivery from its receipt. Historical transcripts do not resume work.
Read review pages and queued work before managing a review page or accepting a queued assignment. Do not create, assign or attach other tasks/workers, and leave manual card moves to the owner/Hydra. Ask Hydra when work needs splitting.
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.

