ClawCode features a unique experience-based learning framework that turns development data into reusable, evolvable knowledge.
Idea → Memory → Plan → Code → Verify → Review → Learned Experience
A function representing the gap between a goal and its outcome. The gap drives improvement.
Dimensions: model_experience, agent_experience, skill_experience, team_experience
A personal/task-level capsule representing an evolvable triplet:
(Instinct, Experience, Skill)
A team collaboration capsule including:
- Collaboration steps and topology
- Handoffs between roles
- Role-level ECAP triplets per team member
flowchart LR
taskExec[Task execution and tool observations] --> expSignal[Experience signal extraction]
expSignal --> ecapCreate[ECAP create and structured store]
ecapCreate --> ecapApply[ECAP retrieve and apply before new work]
ecapApply --> taskOutcome[Outcome and verification]
taskOutcome --> feedback[experience_feedback scores]
feedback --> evolveSkill[instinct_experience_skill evolution]
evolveSkill --> teamCollab[clawteam collaboration]
teamCollab --> tecapCreate[TECAP create or upgrade]
tecapCreate --> teamApply[team_experience_apply context]
teamApply --> loopGate[deep_loop convergence and write-back]
loopGate --> expSignal
/learn # Learn from recent tool observations
/learn-orchestrate # observe → evolve → import-to-skill-store| Command | Purpose |
|---|---|
/experience-create |
Create ECAP from recent observations |
/experience-status |
List available capsules with filters |
/experience-export |
Export as JSON/Markdown |
/experience-import |
Import from file or URL |
/experience-apply |
Apply as one-shot prompt context |
/experience-feedback |
Record success/failure score |
| Command | Purpose |
|---|---|
/team-experience-create |
Create from collaborative traces |
/team-experience-status |
List with team/problem filters |
/team-experience-export |
Export as JSON/Markdown |
/team-experience-import |
Import from file or URL |
/team-experience-apply |
Apply as collaboration context |
/team-experience-feedback |
Record feedback score |
Short aliases: /tecap-* (maps to /team-experience-*)
<data>/
├── learning/
│ ├── experience/
│ │ ├── capsules/ # ECAP capsules
│ │ └── exports/ # Exported capsules
│ ├── team_experience/ # TECAP capsules
│ ├── observations/ # Raw observations
│ └── feedback.jsonl # Feedback scores
└── ...
- Privacy tiers: Redaction, audit trails, feedback scores
- Compatibility flags:
--v1-compatiblefor migration - Quality gates: Capsule validation before application
- Governance metadata:
schema_meta,quality_score,transfer
Model → Agent → Team
Instinct → Experience → Skill
- Instinct: Low-level reusable rules extracted from observations
- Experience: Structured capsules with context and outcomes
- Skill: Evolved skills ready for future task application
/experience-dashboard # ECAP metrics dashboard
/experience-dashboard --json # JSON output
/experience-dashboard --no-alerts # Without alert noise
/closed-loop-contract # Config contract coverage
/instinct-status # Learned instincts by domain
/instinct-export # Export with filters