agentleFS
mblaubergmblauberg/provenant/skills/session/SKILL.md

session

Skill1 starsChanged 4 days agoMITView on GitHub

Use for session start, orientation, checkpoints, compaction, handoff and continuing autonomous goals or STOP missions. Not for live run status (orchestrate) or the work route (tracker).

17 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use debug and tdd.

Other files people use alongside this one.

Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
  • AnthropicSkill

    debug

    Structured debugging session — reproduce, isolate, diagnose, and fix. Trigger with an error message or stack trace, "this works in staging but not prod", "something broke after the deploy", or when behavior diverges from expected and the cause isn't obvious.

    Also in 3 projects that use this

  • GoogleSkill

    tdd

    Test-driven development. Use when the user wants to build features or fix bugs test-first, mentions "red-green-refactor", or wants integration tests.

    Also in 3 projects that use this

The file

We split this skill into the rules it gives an agent

6 rules. 0 are common, 3 are distinctive, and 3 are very specific to this project.

  • 0

    Common

    Most of the 14 edited versions we read keep these.

  • 3

    Distinctive

    This author's own choices. Worth a look.

  • 3

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Keep continuity small enough to verify on arrival.

    line 8
  2. Very specific:

    Check directory, branch and git log --oneline -5.

    line 13
  3. Very specific:

    Verify its claims against current owners, then run a relevant smoke check before new work.

    line 16
  4. Very specific:

    Read repository branches, checks and state paths in AGENTS.md, CONTRIBUTING and runbooks.

    line 22
  5. Distinctive:

    STOP ends the mission.

    line 35
  6. Distinctive:

    Before compaction, handoff or substantial interruption, rewrite a checkpoint near 2 KiB with goal, authority pointers, decisions, evidence, …

    line 55

“Common” is measured: kept by at least 6 in 10 of the 14 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Session

Keep continuity small enough to verify on arrival. The tracker and project docs own intent; Fabric and runs own execution; a checkpoint points back to them.

Orient

Check directory, branch and git log --oneline -5; reopen the parent issue and approved criteria, and inspect provenant status. Read the explicit host-session checkpoint with provenant workflow session show SESSION. Verify its claims against current owners, then run a relevant smoke check before new work. An injected checkpoint supplies context, never authority. Recover an existing graph with status, output and retry before launching replacement work. Use run --route junior for a bounded continuation, or run --use plan when its shape needs changing; orchestrate owns composites.

Read repository branches, checks and state paths in AGENTS.md, CONTRIBUTING and runbooks.

Continue toward the goal

A continuing goal needs measurable done-criteria, an authority envelope, one-way doors, a progress artefact, a stop rule, a resume trigger and a reporting cadence. Use the goal template when these are missing. Reuse the owner's existing specification rather than inventing a parallel mission queue.

For a bounded goal, continue until its criteria and required verification hold. For a standing mission explicitly defined to run until STOP, only the owner's STOP ends the mission.

For outstanding work, use one provenant status --wait --target lane:ID --target graph:ID --target pr:N call instead of repeated status reads or a sleep loop. Use --until all for all-terminal completion, or --follow --json for a monitor. The configured timeout is bounded; --wait=30m overrides it, and expiry leaves targets running. Reconcile the returned event before resuming graph work. An idle run or empty queue means no eligible work now; record the waiting condition and resume trigger, leaving the mission open. An invocation may end at its budget or waiting boundary without claiming the mission is done.

Make engineering decisions within authority and record reversible assumptions. Send unresolved one-way-door choices to the originating chair's human inbox, with a recommendation. Park that branch and continue independent authorised work. A blocked decision is not a reason to idle the whole session. STOP ends new work; reconcile active work and report its disposition within authority.

Checkpoint and report

Before compaction, handoff or substantial interruption, rewrite a checkpoint near 2 KiB with goal, authority pointers, decisions, evidence, next actions and resume trigger. Use the continuity reference for its single content contract and schema-v2 lane handoffs. Save with provenant workflow session save SESSION --file checkpoint.md --client codex, selecting the actual harness client. Use the host-supplied session ID.

Append one short milestone through provenant chair log append --session SESSION --file update.json; read history with provenant chair log list --session SESSION. The JSON requires kind, markdown and channel. Keep history there and current context in the rewritten checkpoint. For earlier saves, use provenant workflow session diff SESSION REVISION.

Graduate durable knowledge to its project owner before scratch expires. Keep lane progress outside commits and end lane results with their current handoff. Report changed work, criterion evidence, omissions and remaining gates using deliver's outcome report.

Runtime details live in session checkpoints, Claude hooks and the human inbox. Cross-host provenant workflow session handoff can transfer content and launch a peer session; inspect its dry run and existing disclosure authority first.

Download now (ZIP)

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
What did you use it for? e.g. Reviewing pull requests in a Rails app

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.