skills.rest
isizono
isizonoCOMMUNITY
Explore all skills by this author →
GitHub

recording

Records discussion logs and session artifacts into a persistent memory database.

3|1|Updated Dec 9, 2025
Search Tags:#logging#SQLite#claude code#decision tracking#session memory#context persistence#artifact storage
One-click install
npx skills add https://github.com/isizono/calm --skill recording-isizono
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: recording
Source: https://github.com/isizono/calm/tree/main/skills/recording
Command: npx skills add https://github.com/isizono/calm --skill recording-isizono

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve? AI coding sessions lose context when they end, so decisions, discussion history, and work artifacts vanish between sessions. This Skill provides the judgment criteria for when and how to persist session events (logs) and deliverables (materials) so later sessions can resume with full context. ## Core Features & Use Cases - Trigger-based log recording: Defines five concrete triggers (L1-L5) for calling add_logs, such as comparing multiple proposals, resolving blockers, handling PR review fixes, policy changes, and bug observations. - Artifact persistence: Defines four triggers (M1-M4) for calling add_material to save drafts, research reports, WIP diffs, and comparison tables as raw data without summarization. - Signal separation: Routes CALM-internal failures, friction, and contradictions to report_signal instead of add_logs, with a detailed taxonomy reference for ambiguous cases. - Use Case: During a long design discussion you compare three architecture options and pick one; the Skill instructs the AI to record the adopted option, rejected options, and rejection reasons via add_logs so the next session inherits the rationale. ## Quick Start Ask the AI to record the discussion outcome and any drafts from this session into persistent memory using the recording guidelines.

Frequently Asked Questions about recording

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I persist AI session context across conversations?▼

Use the recording guidelines to call add_logs for discussion history and add_material for artifacts whenever trigger conditions occur. Records are stored in a SQLite database and automatically surfaced in later sessions through check-in and hybrid search.

When should I record a log versus a material artifact?▼

Logs capture process and reasoning such as proposal comparisons, blocker resolutions, and policy changes. Materials capture static deliverables like drafts, research reports, WIP diffs, and comparison tables, saved as raw data without summarization.

What triggers automatic recording of discussion history?▼

Five triggers apply: comparing three or more proposals, resolving a blocked task, handling multiple PR review fixes, user-directed policy changes, and observing bugs. When in doubt, record rather than skip, since missed context is only recoverable via manual sync-memory.

Does recording require user confirmation before saving?▼

No confirmation is needed for add_material, since artifacts can be saved as soon as they exist, unlike add_decisions which requires mutual agreement. Logs and materials are recorded proactively based on the trigger criteria.

How are tool errors and contradictions reported differently from logs?▼

CALM-internal failures, usability friction, and contradictions with existing records go to report_signal, not add_logs. The L5 bug trigger covers only bugs in the user's target system, keeping internal tool signals separate.

What are the limitations of trigger-based session recording?▼

Trigger-based detection does not guarantee complete coverage of all session events. Missed records are only recovered if the user explicitly runs /sync-memory, so the guideline is to favor recording more rather than less.

Related Skills

Explore More Software Engineering Skills→
swarm-ai-researchswarm-ai-research

Idea Capture

Capture and triage Telegram messages into a structured memory log.

Official
Intermediate
Signet-AISignet-AI
237

recall

Query persistent memories using hybrid vector and keyword search.

Official
Intermediate
dthompson-jtidthompson-jti

audit-docs

Detect documentation drift between knowledge-base content and implemented code.

Community
Intermediate
QredenceQredence
108

initiate-memory

Initialize and reorganize agent memory with a structured project context plan.

Official
Basic