git-intent 0.3.2__tar.gz → 0.5.0__tar.gz

{git_intent-0.3.2/src/git_intent.egg-info → git_intent-0.5.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: git-intent
-Version: 0.3.2
+Version: 0.5.0
 Summary: Semantic history for agent-driven development. Records what you did and why.
 Author: Zeng Deyang
 License-Expression: MIT
@@ -74,13 +74,35 @@ start → snap → done
 ## Example
 
 ```bash
-itt init
+pipx install git-intent
+itt init                    # creates .intent/
 itt start "Fix login timeout"
 itt snap "Increase timeout to 30s" -m "5s too short for slow networks"
 git add . && git commit -m "fix timeout"
 itt done
 ```
 
+## Why Not Just…
+
+| Approach | What it does well | What falls through |
+| --- | --- | --- |
+| **Git commit messages** | Records what changed per commit | No goal structure across commits; rationale is afterthought; no work-in-progress state |
+| **CLAUDE.md / .cursorrules** | Gives agents project-level instructions | Static — doesn't track active tasks, decisions, or progress; must be manually maintained |
+| **TODO comments** | Marks incomplete work in-place | Scattered across files; no lifecycle; no rationale; agents must grep and guess priority |
+| **Notion / Linear / Jira** | Rich project tracking for humans | External to the repo; agents can't read them without API integration; overhead is high for solo/agent workflows |
+| **Agent memory** (e.g. Claude Code memory) | Persists user preferences across sessions | Tied to one platform; not versioned with code; not shareable across agents or teammates |
+| **Ad-hoc context files** (e.g. `context.md`) | Quick, zero-tooling setup | No schema — every project invents its own format; no lifecycle management; grows stale silently |
+
+**Intent occupies a specific gap**: structured, versioned, task-scoped context that lives *in the repo* and works across any agent platform.
+
+- **Structured** — JSON objects with defined schema, not free text an agent must interpret
+- **Task-scoped** — an intent has a lifecycle (`open → done`); snaps are ordered steps, not a pile of notes
+- **Versioned** — `.intent/` is committed alongside code; `git blame` works on your decisions too
+- **Platform-agnostic** — any agent that reads JSON can use it; no vendor lock-in
+- **Minimal** — two objects (intent, snap), one CLI, zero dependencies; adds seconds to a workflow, not minutes
+
+The closest alternative is writing a `context.md` by hand. Intent trades that flexibility for consistency: a schema agents can rely on without per-project prompt engineering.
+
 ## Where This Is Going
 
 `.intent/` is a protocol, not just a tool.
@@ -91,22 +113,36 @@ itt done
 
 ## Install
 
+**Users:**
+
 ```bash
-pip install git-intent
+pipx install git-intent
 ```
 
-Or from source:
+**Configure your agent:** Intent ships a ready-to-use skill file in [`skills/git-intent.md`](skills/git-intent.md). Copy it into your agent's skill directory so it knows the Intent workflow:
+
+```bash
+# Claude Code
+mkdir -p .claude/skills
+curl -o .claude/skills/git-intent.md \
+  https://raw.githubusercontent.com/dozybot001/Intent/main/skills/git-intent.md
+```
+
+For other agents, adapt `skills/git-intent.md` to your platform's instruction format.
+
+**Contributors:**
 
 ```bash
-git clone https://github.com/dozybot001/Intent.git && cd Intent
-pip install -e .
+git clone https://github.com/dozybot001/Intent.git
+cd Intent
+python3 ./itt   # dev entry point, no install needed
 ```
 
 ## Commands
 
 | Command | Purpose |
 | --- | --- |
-| `itt init` | Initialize `.intent/` in a Git repo |
+| `itt init` | Initialize `.intent/` |
 | `itt start <title>` | Open an intent |
 | `itt snap <title> [-m why]` | Record a snap |
 | `itt done` | Close the active intent |
@@ -121,4 +157,4 @@ pip install -e .
 ## Documentation
 
 - [CLI spec](https://github.com/dozybot001/Intent/blob/main/docs/cli.EN.md) — objects, commands, JSON output contract
-- [Agent integration](https://github.com/dozybot001/Intent/blob/main/docs/agent-integration.md) — copy-paste snippets for Claude Code, Cursor, AGENTS.md
+- [Dogfooding](https://github.com/dozybot001/Intent/blob/main/docs/dogfooding.md) — how we built Intent with Intent

{git_intent-0.3.2 → git_intent-0.5.0}/README.md

@@ -50,13 +50,35 @@ start → snap → done
 ## Example
 
 ```bash
-itt init
+pipx install git-intent
+itt init                    # creates .intent/
 itt start "Fix login timeout"
 itt snap "Increase timeout to 30s" -m "5s too short for slow networks"
 git add . && git commit -m "fix timeout"
 itt done
 ```
 
+## Why Not Just…
+
+| Approach | What it does well | What falls through |
+| --- | --- | --- |
+| **Git commit messages** | Records what changed per commit | No goal structure across commits; rationale is afterthought; no work-in-progress state |
+| **CLAUDE.md / .cursorrules** | Gives agents project-level instructions | Static — doesn't track active tasks, decisions, or progress; must be manually maintained |
+| **TODO comments** | Marks incomplete work in-place | Scattered across files; no lifecycle; no rationale; agents must grep and guess priority |
+| **Notion / Linear / Jira** | Rich project tracking for humans | External to the repo; agents can't read them without API integration; overhead is high for solo/agent workflows |
+| **Agent memory** (e.g. Claude Code memory) | Persists user preferences across sessions | Tied to one platform; not versioned with code; not shareable across agents or teammates |
+| **Ad-hoc context files** (e.g. `context.md`) | Quick, zero-tooling setup | No schema — every project invents its own format; no lifecycle management; grows stale silently |
+
+**Intent occupies a specific gap**: structured, versioned, task-scoped context that lives *in the repo* and works across any agent platform.
+
+- **Structured** — JSON objects with defined schema, not free text an agent must interpret
+- **Task-scoped** — an intent has a lifecycle (`open → done`); snaps are ordered steps, not a pile of notes
+- **Versioned** — `.intent/` is committed alongside code; `git blame` works on your decisions too
+- **Platform-agnostic** — any agent that reads JSON can use it; no vendor lock-in
+- **Minimal** — two objects (intent, snap), one CLI, zero dependencies; adds seconds to a workflow, not minutes
+
+The closest alternative is writing a `context.md` by hand. Intent trades that flexibility for consistency: a schema agents can rely on without per-project prompt engineering.
+
 ## Where This Is Going
 
 `.intent/` is a protocol, not just a tool.
@@ -67,22 +89,36 @@ itt done
 
 ## Install
 
+**Users:**
+
 ```bash
-pip install git-intent
+pipx install git-intent
 ```
 
-Or from source:
+**Configure your agent:** Intent ships a ready-to-use skill file in [`skills/git-intent.md`](skills/git-intent.md). Copy it into your agent's skill directory so it knows the Intent workflow:
+
+```bash
+# Claude Code
+mkdir -p .claude/skills
+curl -o .claude/skills/git-intent.md \
+  https://raw.githubusercontent.com/dozybot001/Intent/main/skills/git-intent.md
+```
+
+For other agents, adapt `skills/git-intent.md` to your platform's instruction format.
+
+**Contributors:**
 
 ```bash
-git clone https://github.com/dozybot001/Intent.git && cd Intent
-pip install -e .
+git clone https://github.com/dozybot001/Intent.git
+cd Intent
+python3 ./itt   # dev entry point, no install needed
 ```
 
 ## Commands
 
 | Command | Purpose |
 | --- | --- |
-| `itt init` | Initialize `.intent/` in a Git repo |
+| `itt init` | Initialize `.intent/` |
 | `itt start <title>` | Open an intent |
 | `itt snap <title> [-m why]` | Record a snap |
 | `itt done` | Close the active intent |
@@ -97,4 +133,4 @@ pip install -e .
 ## Documentation
 
 - [CLI spec](https://github.com/dozybot001/Intent/blob/main/docs/cli.EN.md) — objects, commands, JSON output contract
-- [Agent integration](https://github.com/dozybot001/Intent/blob/main/docs/agent-integration.md) — copy-paste snippets for Claude Code, Cursor, AGENTS.md
+- [Dogfooding](https://github.com/dozybot001/Intent/blob/main/docs/dogfooding.md) — how we built Intent with Intent

{git_intent-0.3.2 → git_intent-0.5.0}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "git-intent"
-version = "0.3.2"
+version = "0.5.0"
 description = "Semantic history for agent-driven development. Records what you did and why."
 requires-python = ">=3.9"
 readme = "README.md"

{git_intent-0.3.2 → git_intent-0.5.0/src/git_intent.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: git-intent
-Version: 0.3.2
+Version: 0.5.0
 Summary: Semantic history for agent-driven development. Records what you did and why.
 Author: Zeng Deyang
 License-Expression: MIT
@@ -74,13 +74,35 @@ start → snap → done
 ## Example
 
 ```bash
-itt init
+pipx install git-intent
+itt init                    # creates .intent/
 itt start "Fix login timeout"
 itt snap "Increase timeout to 30s" -m "5s too short for slow networks"
 git add . && git commit -m "fix timeout"
 itt done
 ```
 
+## Why Not Just…
+
+| Approach | What it does well | What falls through |
+| --- | --- | --- |
+| **Git commit messages** | Records what changed per commit | No goal structure across commits; rationale is afterthought; no work-in-progress state |
+| **CLAUDE.md / .cursorrules** | Gives agents project-level instructions | Static — doesn't track active tasks, decisions, or progress; must be manually maintained |
+| **TODO comments** | Marks incomplete work in-place | Scattered across files; no lifecycle; no rationale; agents must grep and guess priority |
+| **Notion / Linear / Jira** | Rich project tracking for humans | External to the repo; agents can't read them without API integration; overhead is high for solo/agent workflows |
+| **Agent memory** (e.g. Claude Code memory) | Persists user preferences across sessions | Tied to one platform; not versioned with code; not shareable across agents or teammates |
+| **Ad-hoc context files** (e.g. `context.md`) | Quick, zero-tooling setup | No schema — every project invents its own format; no lifecycle management; grows stale silently |
+
+**Intent occupies a specific gap**: structured, versioned, task-scoped context that lives *in the repo* and works across any agent platform.
+
+- **Structured** — JSON objects with defined schema, not free text an agent must interpret
+- **Task-scoped** — an intent has a lifecycle (`open → done`); snaps are ordered steps, not a pile of notes
+- **Versioned** — `.intent/` is committed alongside code; `git blame` works on your decisions too
+- **Platform-agnostic** — any agent that reads JSON can use it; no vendor lock-in
+- **Minimal** — two objects (intent, snap), one CLI, zero dependencies; adds seconds to a workflow, not minutes
+
+The closest alternative is writing a `context.md` by hand. Intent trades that flexibility for consistency: a schema agents can rely on without per-project prompt engineering.
+
 ## Where This Is Going
 
 `.intent/` is a protocol, not just a tool.
@@ -91,22 +113,36 @@ itt done
 
 ## Install
 
+**Users:**
+
 ```bash
-pip install git-intent
+pipx install git-intent
 ```
 
-Or from source:
+**Configure your agent:** Intent ships a ready-to-use skill file in [`skills/git-intent.md`](skills/git-intent.md). Copy it into your agent's skill directory so it knows the Intent workflow:
+
+```bash
+# Claude Code
+mkdir -p .claude/skills
+curl -o .claude/skills/git-intent.md \
+  https://raw.githubusercontent.com/dozybot001/Intent/main/skills/git-intent.md
+```
+
+For other agents, adapt `skills/git-intent.md` to your platform's instruction format.
+
+**Contributors:**
 
 ```bash
-git clone https://github.com/dozybot001/Intent.git && cd Intent
-pip install -e .
+git clone https://github.com/dozybot001/Intent.git
+cd Intent
+python3 ./itt   # dev entry point, no install needed
 ```
 
 ## Commands
 
 | Command | Purpose |
 | --- | --- |
-| `itt init` | Initialize `.intent/` in a Git repo |
+| `itt init` | Initialize `.intent/` |
 | `itt start <title>` | Open an intent |
 | `itt snap <title> [-m why]` | Record a snap |
 | `itt done` | Close the active intent |
@@ -121,4 +157,4 @@ pip install -e .
 ## Documentation
 
 - [CLI spec](https://github.com/dozybot001/Intent/blob/main/docs/cli.EN.md) — objects, commands, JSON output contract
-- [Agent integration](https://github.com/dozybot001/Intent/blob/main/docs/agent-integration.md) — copy-paste snippets for Claude Code, Cursor, AGENTS.md
+- [Dogfooding](https://github.com/dozybot001/Intent/blob/main/docs/dogfooding.md) — how we built Intent with Intent

{git_intent-0.3.2 → git_intent-0.5.0}/src/intent_cli/__init__.py

@@ -26,4 +26,4 @@ if __version__ is None:
     try:
         __version__ = metadata.version(PACKAGE_NAME)
     except metadata.PackageNotFoundError:
-        __version__ = "0.3.2"
+        __version__ = "0.4.0"

{git_intent-0.3.2 → git_intent-0.5.0}/src/intent_cli/cli.py

@@ -60,6 +60,7 @@ def build_parser() -> argparse.ArgumentParser:
 
     list_p = sub.add_parser("list", help="List objects")
     list_p.add_argument("type", choices=["intent", "snap"])
+    list_p.add_argument("--intent", dest="intent_id", help="Filter snaps by intent ID")
 
     show_p = sub.add_parser("show", help="Show a single object by ID")
     show_p.add_argument("id")
@@ -130,7 +131,7 @@ def main(argv: Optional[list[str]] = None) -> int:
             return EXIT_SUCCESS
 
         if args.command == "list":
-            items = repo.list_objects(args.type)
+            items = repo.list_objects(args.type, intent_id=getattr(args, "intent_id", None))
             emit(ok("list", items, count=len(items)))
             return EXIT_SUCCESS
 

{git_intent-0.3.2 → git_intent-0.5.0}/src/intent_cli/core.py

@@ -382,7 +382,7 @@ class IntentRepository:
             "warnings": git_warnings,
         }
 
-    def list_objects(self, object_name: str) -> List[Dict[str, Any]]:
+    def list_objects(self, object_name: str, intent_id: Optional[str] = None) -> List[Dict[str, Any]]:
         self.ensure_git()
         self.ensure_initialized()
         if object_name not in ("intent", "snap"):
@@ -392,7 +392,10 @@ class IntentRepository:
                 f"Unknown object type: {object_name}",
                 suggested_fix="Use 'intent' or 'snap'.",
             )
-        return sorted(self.store.list_objects(object_name), key=object_sort_key, reverse=True)
+        items = self.store.list_objects(object_name)
+        if intent_id and object_name == "snap":
+            items = [s for s in items if s.get("intent_id") == intent_id]
+        return sorted(items, key=object_sort_key, reverse=True)
 
     def show_object(self, object_id: str) -> Dict[str, Any]:
         self.ensure_git()

{git_intent-0.3.2 → git_intent-0.5.0}/src/intent_cli/store.py

@@ -51,6 +51,7 @@ class IntentStore:
         }
         write_json(self.config_path, config)
         write_json(self.state_path, state)
+
         return config, state
 
     def load_state(self) -> Dict[str, Any]: