A self-organizing Obsidian vault that gives AI coding agents persistent memory.
TypeScript ai-agentsclaude-codecodex-cligemini-cliknowledge-managementobsidianobsidian-vaultpersistent-memorysecond-braintemplate
fix: match every SessionStart source under Codex and Gemini (#358)
Brenno Ferrari768bad39835c
diff --git a/.claude/scripts/tests/hook-config.test.ts b/.claude/scripts/tests/hook-config.test.tsindex 1cfb04d..9f4d482 100644--- a/.claude/scripts/tests/hook-config.test.ts+++ b/.claude/scripts/tests/hook-config.test.ts@@ -34,6 +34,7 @@ type HookConfig = {readonly hooks?: Record<string,ReadonlyArray<{+ readonly matcher?: string;readonly hooks?: ReadonlyArray<{readonly type?: string;readonly command?: string;@@ -420,3 +421,57 @@ describe("hook config — Codex and Gemini commands run under their agents' shel});}});++describe("hook config — SessionStart fires for every source the agent documents (#352)", () => {+ // Each agent matches SessionStart's matcher against the session's source+ // its own way, and a matcher written for one rule silently matches nothing+ // under another: Gemini's "startup|resume" equalled no source, so+ // session-start never ran there.+ const regex = (matcher: string, source: string) => new RegExp(`^(?:${matcher})$`).test(source);+ // Gemini: a lifecycle matcher is an exact string, "" or "*" matching all+ // (packages/core/src/hooks/hookPlanner.ts: matchesTrigger is+ // matcher === trigger; gemini-cli main as of 2026-10-13).+ const exact = (matcher: string, source: string) => ["", "*", source].includes(matcher.trim());+ const FOUR_SOURCES = ["startup", "resume", "clear", "compact"];+ const agents: ReadonlyArray<{+ readonly label: string;+ readonly path: string;+ readonly sources: readonly string[];+ readonly matches: (matcher: string, source: string) => boolean;+ }> = [+ // Claude Code: a regex over startup|resume|clear|compact (hooks reference).+ { label: "Claude Code", path: ".claude/settings.json", sources: FOUR_SOURCES, matches: regex },+ // Codex: "matcher is applied to source for this event. Values are startup,+ // resume, clear, and compact", and "The matcher field is a regex string"+ // (developers.openai.com/codex/hooks, as of 2026-10-13).+ { label: "Codex CLI", path: ".codex/hooks.json", sources: FOUR_SOURCES, matches: regex },+ // Gemini: sources startup|resume|clear (docs/hooks/reference.md).+ { label: "Gemini CLI", path: ".gemini/settings.json", sources: ["startup", "resume", "clear"], matches: exact },+ ];++ for (const agent of agents) {+ test(`${agent.label}: session-start runs on every documented source`, () => {+ const entries = (loadConfig(agent.path).hooks?.["SessionStart"] ?? []).filter((e) =>+ (e.hooks ?? []).some((h) => h.command?.includes("session-start.ts")),+ );+ assert.ok(entries.length > 0, "a SessionStart entry runs session-start.ts");+ for (const source of agent.sources) {+ const fires = entries.some((e) => e.matcher === undefined || agent.matches(e.matcher, source));+ assert.ok(fires, `no SessionStart matcher fires on source "${source}"`);+ }+ });+ }++ // Gemini's tool events (BeforeTool, AfterTool) take a regex; every other+ // event compares its matcher exactly, so a regex there matches nothing.+ test("Gemini CLI: no lifecycle matcher is written as a regex", () => {+ const TOOL_EVENTS = new Set(["BeforeTool", "AfterTool"]);+ for (const [event, entries] of Object.entries(loadConfig(".gemini/settings.json").hooks ?? {})) {+ if (TOOL_EVENTS.has(event)) continue;+ for (const { matcher } of entries) {+ if (matcher === undefined) continue;+ assert.match(matcher.trim(), /^(\*|[\w-]*)$/, `${event}'s matcher ${JSON.stringify(matcher)} is compared exactly`);+ }+ }+ });+});diff --git a/.codex/hooks.json b/.codex/hooks.jsonindex 81fe1cf..3428b25 100644--- a/.codex/hooks.json+++ b/.codex/hooks.json@@ -2,7 +2,7 @@"hooks": {"SessionStart": [{- "matcher": "startup|resume",+ "matcher": "startup|resume|clear|compact","hooks": [{"type": "command",diff --git a/.gemini/settings.json b/.gemini/settings.jsonindex 9b22ee4..2ed2ae7 100644--- a/.gemini/settings.json+++ b/.gemini/settings.json@@ -7,7 +7,7 @@"hooks": {"SessionStart": [{- "matcher": "startup|resume",+ "matcher": "*","hooks": [{"type": "command",diff --git a/ARCHITECTURE.md b/ARCHITECTURE.mdindex 1847764..111e128 100644--- a/ARCHITECTURE.md+++ b/ARCHITECTURE.md@@ -887,7 +887,7 @@ Step 2 is not documentation garnish. Measured: with the server wired and no repo## Multi-Agent Portability-The same scripts serve three agents. Each agent has its own config file mapping equivalent lifecycle events to the shared scripts. The checklist runs on `Stop` for Claude Code and Codex (per response, deduped) and on `SessionEnd` for Gemini (at shutdown). The script branches on the documented event name and on what a payload field means, never on which agent sent it. Where agents spell one field differently, every spelling is read, so the field can never tell the agents apart: Grok's Stop sends the re-entry flag as `stopHookActive`, and the checklist reads it beside `stop_hook_active` (#323).+The same scripts serve three agents. Each agent has its own config file mapping equivalent lifecycle events to the shared scripts. The checklist runs on `Stop` for Claude Code and Codex (per response, deduped) and on `SessionEnd` for Gemini (at shutdown). The script branches on the documented event name and on what a payload field means, never on which agent sent it. Where agents spell one field differently, every spelling is read, so the field can never tell the agents apart: Grok's Stop sends the re-entry flag as `stopHookActive`, and the checklist reads it beside `stop_hook_active` (#323). Matchers differ the same way. Claude Code and Codex read a SessionStart matcher as a regex over the source (`startup|resume|clear|compact`). Gemini compares a lifecycle matcher to the source as an exact string, with `""` or `"*"` matching all (gemini-cli `hookPlanner.ts`, as of 2026-10-13). A `startup|resume` matcher once matched nothing there, so session-start never ran under Gemini (#352). `tests/hook-config.test.ts` checks every agent's SessionStart matcher under that agent's own rule.Session-boundary output is JSON-or-nothing on all three agents. On the hook path `session-start.ts` prints one `hookSpecificOutput` envelope too, for every agent (#340). `stop-checklist.ts` emits `{}` when a Stop has nothing new to say. What the user sees is `systemMessage`, the one user-facing field all three agents share: on a Stop whose findings changed, a one-line-per-section summary, with the full report saved for the next prompt's `UserPromptSubmit` to hand to the agent unseen. Only when the report cannot be saved does it go out as Stop feedback, which Claude Code prints in full; whether Codex honours Stop feedback is unverified. SessionEnd (Gemini) has no next prompt, so there the full report is `systemMessage` only. `om-wrap-up` still owns the deliberate acting pass at the end of a session.