agentleFS
mystilleefmystilleef/spae-framework/skills/spae-prepare/SKILL.md

spae-prepare

Skill2 starsChanged 15 days agoMITView on GitHub

Preparation Orchestrator for the `SPAE` framework. Spawns the `spec`, `plan`, and `inspect` subagents until the `workstream` reaches `phase: build`.

The file

We split this skill into the rules it gives an agent

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

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈6

    Distinctive · estimate

    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. Very specific:

    Read .spae/current/STATE.json to resolve the current phase.

    line 25
  2. Distinctive:

    Every phase maps to exactly one subagent, spawned the same way regardless of phase:

    line 35
  3. Distinctive:

    After each read, check the exit first.

    line 46
  4. Very specific:

    Read .spae/current/STATE.json.

    line 55
  5. Distinctive:

    Always use the subagent tool for subagent invocation.

    line 67
  6. Distinctive:

    Never run subagents in parallel or concurrently.

    line 80
  7. Very specific:

    Every spawn matched the phase-agent map row for the recorded phase.

    line 94
  8. Distinctive:

    Keep prose terse while ensuring clarity.

    line 101
  9. Distinctive:

    Use lists and sub-lists for long sentences and dense paragraphs.

    line 104

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 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.

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

Prepare (SPAE)

When to use

  • The user wants to run the preparatory phases of the SPAE structured workflow autonomously: spec, plan, and inspect.

Goal

  • Spawn the subagent matching the current STATE.json phase, looped until the workstream reaches phase: build.

Input

Read .spae/current/STATE.json to resolve the current phase. Accept an optional argument, the proposal, during initial execution.

STATE.json

See references/STATE.md for the field reference, directives, and phase snapshots.

Phase-agent map

Every phase maps to exactly one subagent, spawned the same way regardless of phase:

phaseSubagent
specspec
planplan
inspectinspect

Workflow

After each read, check the exit first; halt on any Halt conditions entry after every non-exit read and every spawn.

  1. Resolve Workstream:
    • .spae/current/STATE.json present: proceed to step 2; list an unused proposal argument under Findings.
    • Absent, proposal provided: spawn spec with the proposal; await completion; proceed to step 2.
  2. Dispatch Loop:
    • Read .spae/current/STATE.json; record phase.
    • Exit on phase: "build".
    • Spawn the subagent the map names for phase, no arguments; await completion.
    • Repeat step 2.
  3. Report: emit the result per Result directives:
    • Result: Completed on an exit; Halted when a subagent reports Failed or leaves phase unchanged; Failed on any other Halt conditions entry.

Directives

  • Always use the subagent tool for subagent invocation.
  • Route only from STATE.json phase; trust it over subagent output text.
  • Spawn only the subagents the phase-agent map names.
  • Pass the proposal argument verbatim to the bootstrap spec spawn only; never reuse it on later iterations, and never read or expand file-path arguments.

Constraints

  • Restrict activities to spawning subagents, reading .spae/current/STATE.json, and reporting; never write files, run git operations, activate skills, or prompt the user mid-run.
  • Never run subagents in parallel or concurrently.
  • Halt conditions: halt immediately; report per Report with Halted at and Reason:
    • .spae/current/STATE.json absent at start, with no proposal argument.
    • STATE.json missing at a loop read, or holding unparseable JSON syntax.
    • phase matching no phase-agent map row.
    • phase unchanged since the previous spawn.
    • A subagent Failed result or process failure (crash, timeout, AgentError).

Verification

  • Every spawn matched the phase-agent map row for the recorded phase.
  • The run ended on an exit or a Halt conditions entry.
  • The orchestrator wrote no file.

Result directives

  • Minimum words. Maximum signal.
  • Keep prose terse while ensuring clarity.
  • Optimize prose for agent, token, and context efficiency.
  • Split actions, findings, and summaries into terse bullet points.
  • Use lists and sub-lists for long sentences and dense paragraphs.
  • Emit the result template as live markdown—never in a code fence.
  • Output nothing outside the template.

Result template

<!-- prettier-ignore-start -->
### Execution Summary

- **Actions**:
  - [Terse list of actions from spawned subagents]
- **Files**:
  - [Terse list of files affected by spawned subagents]
- **Findings**:
  - [Terse list of findings from spawned subagents]
- **Summary**:
  - [Terse list of summary from spawned subagents]

> **`SPAE` Prepare** • `[workstream]`
> **Result**: [Completed | Halted | Failed]
> **Halted at**: `[Phase]`  _(unless completed)_
> **Reason**: [Terse explanation]     _(unless completed)_
> **Impact**: [Terse impact statement]
<!-- prettier-ignore-end -->

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.