Chronicle Copilot Session Analyst
Turn Copilot session history into standups, usage advice, and searchable records.
The skill explicitly uses a local session database, describes queries as read-only, and forbids fabricated results. However, session text, file paths, and optional cloud synchronization may involve sensitive data; the skill does not define data minimization, redaction, user confirmation, permission boundaries, or rollback/recovery for deletion or reindex failures. Reindexing may also create external cloud effects, so trust is deducted.
Prerequisites, unavailable-tool feedback, local SQLite versus cloud DuckDB dialects, key tables, and several empty-result paths are documented, making the happy path plausible. Reliability is reduced because copilot_sessionStoreSql, GitHub CLI/MCP, and referenced auxiliary skills are not defined here; the supplied reindex section is truncated, and SQL, permission, and sync failures lack sufficiently diagnosable recovery guidance. Static calibration caps this at 10, so the score is 7.
Triggers and scenarios for standups, tips, cost tips, improve, search, and reindex are fairly specific, with local/cloud distinctions and some scope boundaries. Audience boundaries, false-trigger avoidance, version compatibility, and Chinese-language output support are not fully specified. Cloud reachability is also unaddressed, although the core function is not entirely dependent on an overseas service, so the deduction is limited.
The document has useful progressive sections covering prerequisites, actions, query rules, output formats, empty results, and limitations. Repository materials provide MIT licensing, Microsoft attribution, and governance context. The skill itself lacks a version, changelog, named maintenance responsibility, and explicit update path; referenced skills and external tool dependencies are also not fully documented for installation or compatibility, so points are deducted.
The skill covers session summaries, search, usage and cost analysis, and index maintenance, with structured output and anti-fabrication rules that could produce directly usable results. Effectiveness is limited by dependence on indexing, correct backend dialects, external GitHub status checks, sufficient history, and auxiliary skills for some workflows. No representative execution evidence is included; under static calibration the score remains below 7, at 6.
Concrete SQL predicates, schema names, field names, backend differences, output templates, and query safeguards provide limited auditability. There are no skill-specific tests, fixture data, CI coverage, or third-party execution results in the supplied material, and some dependencies lack implementation evidence; the static score is therefore 3.
- Before execution, disclose which session fields will be read and minimize or redact session text, file paths, and PR data.
- Reindex may rebuild local state and synchronize to the cloud; it should require explicit confirmation and document scope, failure feedback, and recovery.
- The tips, improve, and some PR-status workflows depend on tools or skills not defined in this skill; deployment should verify those dependencies.
- The supplied SKILL.md ends partway through the reindex workflow, so complete maintenance, deletion, and failure behavior cannot be verified.
What does this skill do, and when should you use it?
Chronicle is an Agent Skill in the VS Code repository for analyzing Copilot session history. It uses the copilot_sessionStoreSql tool to query sessions, turns, files, references, and checkpoints from local SQLite or optionally synchronized cloud data. It supports standup reports, usage analysis, historical session search, index rebuilding, and guidance for deleting synced session data. It requires github.copilot.chat.localIndex.enabled to be true, and the repository as a whole is MIT licensed.
Calls copilot_sessionStoreSql for read-only SQL queries over sessions, turns, session_files, session_refs, and checkpoints; builds a last-24-hours standup grouped by branch or work stream and checks the current status of referenced pull requests; analyzes recent prompts, tools, files, and workflow patterns to produce 3-5 actionable recommendations; analyzes model and event-level token usage when cloud data is available, or uses local proxies such as session length, repeated file reads, and tool-call patterns; searches historical sessions across summaries, conversation content, file paths, and pull request, issue, or commit references; runs the reindex action to rebuild the local index and synchronize with the cloud; and directs users to the Delete Session Sync Data command for deletion.
- A developer wants a standup based on Copilot activity from the previous 24 hours.
- A developer needs to find an earlier Copilot session by a keyword, file path, or pull request number.
- A user wants workflow recommendations grounded in actual prompting, tool-use, and file-access patterns.
- A VS Code Chat user wants cost guidance based on tokens, model choice, long sessions, or repeated reads.
- A user with an incomplete or newly enabled index needs to rebuild local session data and refresh cloud synchronization.
- A user wants to remove historical session records and needs the supported deletion procedure.
What are this skill's strengths and limitations?
- Covers standups, usage analysis, historical search, and session-store maintenance in one workflow.
- Does not require a pull request for standup inclusion; PR-less sessions remain candidate work items.
- Searches summaries, conversation content, file paths, and pull request, issue, and commit references.
- Distinguishes local SQLite from cloud DuckDB and adjusts its analysis to the available backend.
- Requires github.copilot.chat.localIndex.enabled and an available copilot_sessionStoreSql tool.
- Local SQLite does not record per-event token usage; precise token-level cost analysis requires chat.sessionSync.enabled.
- The supplied material does not document cloud data retention, privacy, or permission details.
- No standalone installation command, test suite, or cross-platform validation is documented in the supplied material.
How do you install this skill?
Chronicle is located at https://github.com/microsoft/vscode in extensions/copilot/assets/prompts/skills/chronicle/SKILL.md. The supplied material does not document a standalone installation command or complete installation procedure; it is part of a monorepo collection containing 61 skills. Before adopting it, confirm that the host can provide copilot_sessionStoreSql and set github.copilot.chat.localIndex.enabled to true in VS Code Settings.
How do you use this skill?
In a host that supports the skill, make requests such as “/chronicle standup”, “/chronicle search authentication”, or “/chronicle reindex”. Use “/chronicle cost-tips” for cost analysis and “/chronicle improve” for instruction recommendations. If copilot_sessionStoreSql is unavailable, enable github.copilot.chat.localIndex.enabled first. To delete data, run Delete Session Sync Data from the Command Palette.