Steer agents to the tracedecay CLI when MCP transport fails

[ScriptedAlchemy]

Summary

• Add a using-the-cli skill to both plugin bundles (byte-identical for the Codex parity test) covering CLI discovery (tracedecay tool lists tools, tracedecay tool <name> --help shows parameters), invocation flags (--json, --args, --project, @file values), when to switch, and guardrails against querying .tracedecay/*.db directly or abandoning tracedecay on transport failure.
• Register the skill in EMBEDDED_PLUGIN_FILES, CODEX_EMBEDDED_PLUGIN_FILES, and the CURSOR_PLUGIN_SKILLS session-steering index.
• Mention the CLI fallback in the always-applied Cursor rule, the Codex session steering context, the MCP serverInstructions, both plugin READMEs, and both plugin manifest descriptions.
• Attach a cli_fallback data field to the generic MCP tool-execution error response so a failing call itself points at the equivalent CLI command.

Motivation: past-session transcripts show agents falling back to raw SQLite queries against .tracedecay/.tokensave databases or dropping tracedecay entirely when the MCP server errors or times out, even though every MCP tool is also exposed as tracedecay tool <name> with built-in help discovery.

Test plan

• cargo check --all-targets clean
• cargo clippy --all-targets clean
• cargo test --lib agents:: (135 passed — bundle coverage, Codex skill parity, skill-index sync)
• cargo test --lib hooks:: (23 passed)
• cargo test --test plugin_skill_contract_test --test hooks_test --test mcp_server_test --test mcp_handler_test — all pass except two pre-existing flaky mcp_handler_test cases (hermes_skill_bridge_mcp_tool_reads_host_owned_profile_state, message_search_reads_profile_sharded_session_db) that race on env vars under full parallelism and pass in isolation

[changeset-bot]

⚠️ No Changeset found

Latest commit: 03216eb

Merging this PR will not cause a version bump for any packages. If these changes should not result in a new version, you're good to go. If these changes should result in a version bump, you need to add a changeset.

Click here to learn what changesets are, and how to add one.

Click here if you're a maintainer who wants to add a changeset to this PR