session-intent-analyzer

name	session-intent-analyzer
description	Update the per-session intent ledger for every user input and provide current intent as semantic context consumed by gates, skill-evolution, and jailbreak-detector. Keep intent summaries without storing raw prompts.
compatibility	["Python 3.11+ standard library","hook-capable environments plus hookless/manual fallback"]
calls	["soft:jailbreak-detector","soft:skill-evolution"]

session-intent-analyzer

session-intent-analyzer maintains a small per-session ledger of the user's current goal, constraints, decisions, non-goals, open questions, and acceptance criteria. It lets skill-evolution interpret tool sequences with intent context and lets jailbreak-detector compare current requests against accumulated session intent.

Contents

• Storage Contract
• When To Use
• Procedure
• Consumer Snapshot
• Hookless Fallback
• Warnings

Storage Contract

• The ledger lives under .tmp/session-intent/<platform>/<session-id>/ in the Ghost-ALICE repo root.
• The installer passes --root <repo>/.tmp/session-intent to hook commands. Manual runs may override the root with GHOST_ALICE_SESSION_INTENT_ROOT.
• intent-state.json stores the latest session intent state.
• Hooks record digest-only observations and intake status. Agents add semantic deltas only when goals, constraints, decisions, or acceptance criteria materially change.
• Scalar intent fields are replaced by the latest semantic delta. Lists such as constraints, non-goals, open questions, criteria, and decisions are deduped or merged by id.
• conduct_feedback records compressed behavioral corrections, how the agent operated versus what the user asked, and merges by id while preserving occurrence_count. Repeated same-id corrections in one session increment occurrence_count; status-only updates do not. session-intent-analyzer captures it and skill-evolution consumes it, so the behavioral-correction loop is complete only across both skills, not in skill-evolution alone.
• model_security_decision is owned by jailbreak-detector. Intake preserves it and must not clobber it.
• Only a current-lineage block decision is carried to downstream-gates.json. Non-block decisions and non-current-lineage decisions are not carried.
• intent-events.jsonl records input observations and intent updates.
• security-events.jsonl records security judgments.
• ../current-session.json points to the current platform session and state path.
• Raw prompts, full conversations, tool output, and secret values are never stored.
• User input is stored only as digest and length.
• Detailed schema lives in references/ledger-schema.md.

When To Use

• Immediately after every user input, before task-router.
• When the user changes goals, constraints, priority, or acceptance criteria.
• When passing --intent-ledger to skill-evolution.
• When jailbreak-detector needs current intent context.

Procedure

1. Resolve platform and session id in this order: explicit session id, hook payload session fields, GHOST_ALICE_SESSION_ID, current-session.json, then unknown.
2. For hook observation, store only input_digest, input_char_count, intake_status=observed, and intent_delta_status=not-provided.
3. Add a semantic delta only when current_goal, user_intent_summary, constraints, non_goals, decisions, open_questions, or acceptance_criteria materially changes. When a completion, recommendation, or choice is anticipated, record verifiable acceptance_criteria from user intent so the final [completion-check] can carry them into acceptance-criteria and bind each claim in claim-evidence-map.
4. Keep corrective lessons general. Do not store long episode details. Preserve the reusable reasoning pattern, not case detail. When the user corrects the agent's conduct (under-delivery, silent scope narrowing, reporting or asking instead of executing, punting a decision the content could resolve, an unrequested trace), record it as a compressed conduct_feedback entry with a stable id, a reusable summary or corrective_rule, and an occurrence_count when known. This is the behavioral-correction signal skill-evolution consumes; the two skills close the loop together, and neither does it alone. Judge "is this a correction" by the same basis the rest of the system uses, state vs reference: a mismatch the input asserts between the agent's prior action or claim and the ledger's accumulated current_goal, constraints, non_goals, decisions, and acceptance_criteria. This is the input-vs-accumulated comparison jailbreak-detector applies for security, turned on the agent's own conduct, not keyword matching. Record proactively too: when you notice a gap between the user's stated intent or constraints and how a skill currently behaves, log it as an open conduct_feedback recommendation with source: inferred. These accumulate as a backlog and are applied only when the user explicitly asks to update.
5. Use consumer_hints when downstream gates need immediate caution or completion criteria.
6. Use scripts/session_intent_ledger.py to update intent-state.json and intent-events.jsonl.
7. If security signals exist, pass intent_summary and intent-state.json to jailbreak-detector.
8. For repeated-action analysis, pass --intent-ledger <intent-state.json> to skill-evolution.

Digest-only hook observation is enough for intake completion. Semantic delta is required only when intent materially changes.

If no delta is needed, leave last_semantic_delta_status=not-provided and mark session-intent-analyzer: done. If a delta was needed but not recorded, mark it hook-observed.

Consumer Snapshot

Snapshots use snake_case acceptance_criteria. Final [completion-check] surfaces convert that to the English control field acceptance-criteria.

{
  "schema_version": "session-intent-ledger.v1",
  "current_goal": "current goal",
  "user_intent_summary": "compressed intent summary",
  "constraints": ["constraint"],
  "non_goals": ["non-goal"],
  "open_questions": ["open question"],
  "acceptance_criteria": [
    {
      "id": "criterion-id",
      "summary": "verifiable completion condition",
      "source": "user-explicit | inferred | previous-tool | system-doc"
    }
  ],
  "decision_count": 1,
  "risk_flags": ["jailbreak-suspected"],
  "consumer_hints": {
    "skill_evolution": ["interpret tool sequence with intent context"]
  },
  "intake_status": "observed",
  "last_semantic_delta_status": "not-provided",
  "semantic_delta_policy": "agent-updates-when-intent-materially-changes"
}

Hookless Fallback

Without hooks, apply the same storage contract before the first response. If the session id cannot be resolved, write degraded state under unknown. Missing ledger context should fall back gracefully; it is not a denial reason by itself.

Warnings

• This skill is not long-term memory.
• Do not promote memory without user approval.
• This skill does not make the final jailbreak decision.
• skill-evolution may use this ledger only as report context.
• Ledger text must be compressed intent, not raw quotes, secrets, system messages, or tool output.