swmcc / swmcc/agentic-development
Agent souls: identity, memory and taste for the factory
Nobody has claimed this yet.
- Dominant language
- Python
- Stars
- 0
- Forks
- 0
- Avg merge
- 2m
- Merged PRs (30d)
- 1
Description
Umbrella and decision record for the souls work. The full design lives in the vault note agent-souls.md (mazarbul, hubs/projects/agentic-development): a soul decomposes into taste (falsifiable convictions), memory (per-repo lessons plus per-character memory), voice and a track record. Thrawn's real characters are the pipeline stages: planner as tactician, integrator as fixer and a new pre-ship skeptic. Executors stay deliberately soul-light.
The #14 runner trial accidentally proved the premise: the discipline gap between codex and pi was a personality gap. codex ships with baked-in convictions, pi ships with none and committed two lint-failing branches because nobody told it to check its work. Souls put those convictions in one versioned place and inject them into whichever harness runs the pane.
Constraint: phase 2 of #14 keeps running throughout. Nothing here may change the planner, executor or swarm behaviour mid-trial except the check-step ticket, which is a declared phase 2 prerequisite.
Task list is maintained in a comment below once the sub-tickets exist.
Contributor guide
No contributing guide indexed for this repository
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
Research direction
Start with the vault note agent-souls.md and issue #14's runner trial to understand the proposed soul components and phase-2 constraint. Map the design into explicit sub-tickets, preserving the existing planner, executor, and swarm behavior except for the declared check-step prerequisite; done means the scope and decisions are recorded without mid-trial changes.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- python
- Domain
- ai
- Issue type
- Feature
- Difficulty
- 5/5
- Estimated time
- Over a week
- Activity status
- Active
- Clarity
- Needs clarification
- Newbie friendliness
- 25/100