Add Oh My Pi source support and Herdr resume scoping (#89)

* Add Oh My Pi session support

* Add distinct Oh My Pi source support

* Add OMP source support and Herdr resume scoping

* Fix Oh My Pi discovery and skill setup

* Consolidate shared agent skill installation

* Fix OMP agent root test fixture

---------

Co-authored-by: emo-nm <james@gs.xyz>
Co-authored-by: Nico Ritschel <ritschel@gmail.com>

Author: 3 people

README.md

@@ -1,6 +1,6 @@
 # memex
 
-Fast local history search for Claude, Codex CLI, Cursor, OpenCode, Pi Coding Agent, OpenClaw, GitHub Copilot CLI, and Hermes usage records. Uses BM-25 and optionally embeds your transcripts locally for hybrid search.
+Fast local history search for Claude, Codex CLI, Cursor, OpenCode, Pi Coding Agent, Oh My Pi, OpenClaw, GitHub Copilot CLI, and Hermes usage records. Uses BM-25 and optionally embeds your transcripts locally for hybrid search.
 
 Mostly intended for agents to use via skill. The intended workflow is to ask agent about a previous session & then the agent can narrow things down & retrieve history as needed.
 
@@ -109,7 +109,11 @@ Then run setup to install the skills:
 memex setup
 ```
 
-Restart Claude, Codex, OpenCode, or Pi after setup.
+The shared `memex-search` skill is installed once at
+`~/.agents/skills/memex-search/SKILL.md` for Codex, OpenCode, Pi, and Oh My Pi.
+Claude Code uses `~/.claude/skills/memex-search/SKILL.md`.
+
+Restart Claude, Codex, OpenCode, Pi, or Oh My Pi after setup.
 
 ## Quickstart
 
@@ -118,6 +122,9 @@ Index (incremental):
 memex index
 ```
 
+The default scan indexes Pi sessions from `~/.pi/agent/sessions` and Oh My Pi sessions
+separately from `~/.omp/agent/sessions` plus named profile session directories.
+
 Plaintext reasoning is excluded by default because it is usually low-value search noise. Opt
 in with `memex index --include-reasoning`; reasoning records remain BM25-only. Encrypted and
 redacted payloads, along with reasoning signature fields, are always excluded.
@@ -210,7 +217,7 @@ Token tracking is disabled by default because it scans and caches local agent lo
 token_usage = true
 ```
 
-Then reconstruct historical token usage from local Claude Code, Codex, Cursor, OpenCode, Pi, OpenClaw, Copilot, and Hermes records:
+Then reconstruct historical token usage from local Claude Code, Codex, Cursor, OpenCode, Pi, Oh My Pi, OpenClaw, Copilot, and Hermes records:
 
 ```
 memex usage
@@ -252,7 +259,7 @@ If you built from source, run setup to install:
 memex setup
 ```
 
-This detects which tools are installed (Claude/Codex/OpenCode/Pi) and presents an interactive menu to select which to configure.
+This detects which tools are installed (Claude/Codex/OpenCode/Pi/Oh My Pi) and presents an interactive menu to select which to configure.
 ## Search modes
 
 | Need | Command |
@@ -267,7 +274,7 @@ This detects which tools are installed (Claude/Codex/OpenCode/Pi) and presents a
 - `--role <user|assistant|tool_use|tool_result>`
 - `--tool <tool_name>`
 - `--session <session_id>`
-- `--source claude|codex|cursor|opencode|pi|openclaw|copilot|hermes`
+- `--source claude|codex|cursor|opencode|pi|omp|openclaw|copilot|hermes`
 - `--since <iso|unix>` / `--until <iso|unix>`
 - `--limit <n>`
 - `--min-score <float>`

fixtures/trajectory_parity/omp.jsonl

@@ -0,0 +1,6 @@
+{"type":"title","v":1,"title":"Inspect the project","source":"auto","updatedAt":"2026-08-13T00:00:03.000Z"}
+{"type":"session","version":3,"id":"omp-session","timestamp":"2026-08-13T00:00:00.000Z","cwd":"/workspace/omp-project","title":"Inspect the project","titleSource":"auto"}
+{"type":"model_change","id":"model-1","parentId":null,"timestamp":"2026-08-13T00:00:00.100Z","model":"openai-codex/gpt-5.6-luna"}
+{"type":"message","id":"user-1","parentId":"model-1","timestamp":"2026-08-13T00:00:01.000Z","message":{"role":"user","content":[{"type":"text","text":"Inspect the project"}],"timestamp":1786665601000}}
+{"type":"message","id":"assistant-1","parentId":"user-1","timestamp":"2026-08-13T00:00:02.000Z","message":{"role":"assistant","content":[{"type":"thinking","thinking":"I will inspect the project."},{"type":"text","text":"I will inspect the project."},{"type":"toolCall","id":"tool-1","name":"Read","arguments":{"path":"README.md"}}],"timestamp":1786665602000,"usage":{"input":12,"output":8,"cacheRead":2,"cacheWrite":1},"model":"gpt-5.6-luna","provider":"openai"}}
+{"type":"message","id":"tool-result-1","parentId":"assistant-1","timestamp":"2026-08-13T00:00:03.000Z","message":{"role":"toolResult","toolCallId":"tool-1","toolName":"Read","content":[{"type":"text","text":"project contents"}],"isError":false,"timestamp":1786665603000}}

herdr-plugin.toml

@@ -1,8 +1,8 @@
 # memex as a herdr plugin: a session desk for your agent history.
 #
 # memex indexes the transcripts every coding agent on this machine leaves behind (Claude Code,
-# Codex, Cursor, OpenCode, Pi, OpenClaw, Copilot). Inside herdr it becomes a native pane: browse
-# and search past sessions, then resume one straight into a new herdr tab.
+# Codex, Cursor, OpenCode, Pi, Oh My Pi, OpenClaw, Copilot). Inside herdr it becomes a native
+# pane: browse and search past sessions, then resume one straight into a new herdr tab.
 #
 # The version here must match Cargo.toml — herdr/install.sh downloads the release tarball tagged
 # with exactly this version.
@@ -11,7 +11,7 @@ name = "memex"
 version = "0.10.1"
 min_herdr_version = "0.7.0"
 platforms = ["macos", "linux"]
-description = "Session desk for your agent history: search every past Claude Code / Codex / Cursor session and resume one into a herdr tab."
+description = "Session desk for your agent history: search past Claude Code / Codex / Cursor / Pi / Oh My Pi sessions and resume one into a herdr tab."
 
 # On `herdr plugin install`, put a working memex binary under $HERDR_PLUGIN_ROOT/bin: reuse one
 # already on PATH if it speaks the herdr surface, else download the matching release tarball, else

herdr/plugin.sh

@@ -203,7 +203,7 @@ resume-last)
     refuse "cannot read pane context ($ctx_state); refusing to resume without directory scoping"
   # memex resolves the most recent resumable session and opens it in a new herdr tab itself,
   # driving the herdr CLI with the HERDR_* env this action already carries.
-  args=(herdr resume-last)
+  args=(herdr resume-last --strict-cwd)
   [ -n "$ctx_cwd" ] && args+=(--cwd "$ctx_cwd")
   # MEMEX_ROOT (matching memex-pane.sh) points every memex call at an alternate data directory.
   [ -n "${MEMEX_ROOT:-}" ] && args+=(--root "$MEMEX_ROOT")

skills/codex/memex-search/SKILL.md

@@ -1,6 +1,6 @@
 ---
 name: memex-search
-description: Search, filter, and retrieve Claude/Codex/Cursor/OpenCode/Pi/Copilot history indexed by the memex CLI. Use when the user wants to index history, run lexical/semantic/hybrid search, fetch full transcripts, or produce LLM-friendly JSON output for RAG.
+description: Search, filter, and retrieve Claude/Codex/Cursor/OpenCode/Pi/Oh My Pi/Copilot history indexed by the memex CLI. Use when the user wants to index history, run lexical/semantic/hybrid search, fetch full transcripts, or produce LLM-friendly JSON output for RAG.
 allowed-tools: Bash(memex:*)
 ---
 
@@ -27,6 +27,7 @@ Use this skill to index local history and retrieve results in a structured, LLM-
   - `--codex/--no-codex` to include or skip Codex logs
   - `--opencode/--no-opencode` to include or skip OpenCode logs
   - `--pi/--no-pi` to include or skip Pi logs
+  - `--omp/--no-omp` to include or skip Oh My Pi logs
   - `--copilot/--no-copilot` to include or skip GitHub Copilot CLI logs
   - `--model <minilm|bge|nomic|gemma|potion>` to select embedding model
   - `--root <path>` to change data root (default: `~/.memex`)