Knowledge Activation Productization Plan¶
Date: 2026-04-01
Status: proposed implementation plan
Source: local prototype validated in /Users/fullerbt/.agents and /Users/fullerbt/.codex/plugins/cache/agentops-marketplace/agentops/local/skills-codex/knowledge-activation/
Problem¶
AgentOps already has strong flywheel primitives, but it is still better at capturing and maintaining knowledge than operationalizing a mature .agents corpus into task-time advantage.
The local prototype exposed the missing product gap:
- packetization and mining are not enough
- users need promoted operator surfaces, not just retrieval substrate
compileis a maintenance skill, not a corpus activation skill
The result is a common failure mode:
- knowledge exists
- knowledge is searchable
- knowledge still does not materially change how future agent sessions begin and execute
Product Decision¶
Introduce a new outer-loop capability in AgentOps.
Working name:
knowledge-activation
The name can change later, but the contract should not.
compile remains the hygiene skill for mine, validate, and defrag.
The new capability owns:
- corpus consolidation
- belief promotion
- playbook generation
- briefing compilation
- gap and thin-topic feedback into the flywheel
What The Prototype Already Proved¶
The local prototype validated this end-to-end shape:
- evidence packet families
- belief book generation
- playbook candidate generation
- per-goal briefing generation
- thin-topic caution handling
Validation already passed locally:
- Python compilation
- unit tests
- skill bundle validation
- full wrapper execution
Local outputs worth treating as seed material:
/Users/fullerbt/.agents/knowledge/book-of-beliefs.md/Users/fullerbt/.agents/playbooks/index.md/Users/fullerbt/.agents/briefings/2026-04-01-turn-agents-into-usable-information-and-spin-the-knowledge-f.md/Users/fullerbt/.agents/knowledge/agent-interaction-operationalization.md/Users/fullerbt/.agents/retros/2026-04-01-knowledge-activation-consolidated-postmortem.md
What AgentOps Should Do Better¶
1. Prefer briefings over giant startup dumps¶
AgentOps currently has strong lifecycle and injection surfaces, but it should more explicitly support:
- a small goal-time briefing
- selected principles
- selected playbooks
- warnings and trust boundaries
- evidence links behind the scenes
This is a better startup surface than broad generic startup context.
2. Separate storage from operator surfaces¶
The product should distinguish:
- packet/chunk/evidence substrate
- belief and playbook promotion layer
- task-time briefing layer
Without this separation, users get organization without activation.
3. Encode trust boundaries¶
Thin topics must be first-class product state, not an implicit caveat buried in docs.
The system should know:
- what is healthy enough to promote
- what stays discovery-only
- what should trigger another mining or review pass
4. Improve prompt-surface guidance¶
AgentOps should give a clear rule:
- stable beliefs belong in
AGENTS.md,CLAUDE.md, and memory - dynamic guidance belongs in a briefing
- repeated failures belong in planning rules, checks, or skill logic
Proposed CLI Surface¶
Add an ao knowledge family:
ao knowledge activate¶
Runs the full outer loop:
- preflight
- evidence consolidation
- belief and playbook promotion
- briefing refresh
- gap reporting
ao knowledge beliefs¶
Builds or refreshes the belief book from promoted evidence.
ao knowledge playbooks¶
Builds or refreshes playbook candidates from healthy topics.
ao knowledge brief --goal "<goal>"¶
Compiles a runtime briefing using:
- relevant beliefs
- relevant playbooks
- warnings
- evidence links
ao knowledge gaps¶
Reports:
- thin topics
- promotion gaps
- weak claims needing review
- next recommended mining work
Proposed Skill Surface¶
Add a new user-invocable skill in the AgentOps repo:
knowledge-activation
The skill should orchestrate the ao knowledge commands and explain the contract.
It should not own the heavy lifting forever. The builders belong in stable product or CLI surfaces, while the skill owns:
- user intent routing
- execution order
- output interpretation
- next-step recommendations
Proposed Output Surfaces¶
The product should standardize these outputs:
.agents/packets/.agents/knowledge/book-of-beliefs.md.agents/playbooks/.agents/briefings/.agents/retros/
Required trust rules:
- packet and chunk layers are substrate
- beliefs, playbooks, and briefings are consumer surfaces
- thin topics remain discovery-only until promoted health improves
Runtime Integration¶
The lifecycle should treat briefings as a first-class startup aid.
Recommended behavior:
ao codex startchecks for an active task or handoff goal- if a recent matching briefing exists, surface it
- if no briefing exists, suggest or build one
- keep the startup surface small and citation-backed
This keeps startup context bounded and aligned with the belief system.
Implementation Waves¶
Wave 1: Upstream the skill contract¶
- Add
crew/nami/skills/knowledge-activation/ - Port the validated local
SKILL.md - Add DAG and output-surface references
- Add a bundle validator
Wave 2: Productize builders¶
- Decide whether to port the current Python builders into
aodirectly or wrap them first - Expose the command family:
ao knowledge activateao knowledge beliefsao knowledge playbooksao knowledge briefao knowledge gaps- Preserve deterministic behavior on unchanged inputs
Wave 3: Runtime integration¶
- Teach the Codex lifecycle to prefer briefings for startup
- Keep prompt surfaces small
- Record trust warnings for thin topics and weak claims
Wave 4: Review and hardening¶
- Add tests for deterministic output
- Add tests for thin-topic handling
- Add docs describing the operator layers
- Run a manual review pass on belief and playbook sharpness
Acceptance Criteria¶
- A user can point AgentOps at a mature
.agentscorpus and get operational outputs, not just mined evidence. - AgentOps produces a belief book, playbook candidates, and a goal-time briefing.
- Thin topics are surfaced explicitly and never silently promoted as canonical truth.
- The new capability complements
compileinstead of bloating it. - Startup interactions improve because the system prefers briefings over broad context dumps.
Risks¶
Retrieval substrate instability¶
If cass index remains unreliable, refresh confidence stays weaker than it should be.
Weak claim promotion¶
Artifact-title leakage into beliefs or playbooks will undermine user trust if not reviewed.
Doc graveyard risk¶
If outputs are generated without clear consumers, the system will create another organized archive instead of a working outer loop.
Recommendation¶
Implement this as a new AgentOps outer-loop capability now.
Do not try to stretch compile until it absorbs this contract. That would blur maintenance and activation and make both harder to reason about.
The winning model is:
compilefor hygieneknowledge-activationfor corpus operationalizationao knowledge briefas the task-time bridge into actual agent behavior