> For the complete documentation index, see [llms.txt](https://docs.algenta.ai/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.algenta.ai/mcp-tools/repository-intelligence.md). # Repository intelligence tools These tools drive Algenta's repository-intelligence lane over MCP: take an immutable snapshot of a saved repository connector, triage it into a bounded evidence bundle, plan a fix, query the dependency graph, simulate patch risk, and apply the result as a patch, a local branch, or a pull request. Every tool is a thin, one-to-one wrapper over an authenticated `/v1/repositories` endpoint, so anything you do here you can reproduce with `curl`. Before you start, install and configure `algenta-mcp` so the `algenta` tools appear in your client — see [MCP server](/sdks/mcp.md). Most tools need a `$REPOSITORY_ID` from a saved repository connector; snapshot, plan, and simulate steps then hand you a `$SNAPSHOT_ID`, `$DECISION_PLAN_ID`, and `$SIMULATION_ID` to chain forward. {% hint style="info" %} Prefer `run_repository_pipeline` for the common path: it runs snapshot → triage → plan → simulate in one call and returns the canonical repository envelope. Reach for the individual tools when you need to inspect or override a single stage. {% endhint %} ## Tools | Tool | What it does | Key inputs | | ------------------------------------------ | --------------------------------------------------------------------------------------------------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | `get_repository_intelligence_capabilities` | List globally supported languages and their ranked support levels. | *(none)* | | `create_repository_snapshot` | Create or reuse an immutable snapshot for a saved repository connector. | `repository_id` (required); `ref`, `include_patterns`, `exclude_patterns`, `max_files`, `max_file_size_bytes` | | `get_repository_snapshot` | Fetch one immutable snapshot by id. | `repository_id`, `snapshot_id` (both required) | | `triage_repository` | Triage a snapshot into a bounded workspace evidence bundle of suspect files and symbols. | `repository_id`, `snapshot_id`, `signals` (all required); `max_evidence_items`, `max_snippet_lines`, `token_budget` | | `create_repository_decision_plan` | Create one immutable DecisionPlan revision from an evidence bundle (resolves `snapshot_id` from triage when omitted). | `repository_id`, `workspace_evidence_bundle_ref` (required); `snapshot_id`, `model` | | `query_repository_graph` | Query a snapshot for dependency, dependent, and change-risk graph edges. | `repository_id` (required) + `snapshot_id` **or** `workspace_evidence_bundle_ref`; `file_path`, `symbol_name`, `direction` (`inbound`\|`outbound`\|`both`), `max_depth` (1–6), `max_nodes` (1–1024) | | `simulate_repository` | Simulate patch risk and return the gated DecisionEnvelope (resolves `snapshot_id` from the plan when omitted). | `repository_id`, `decision_plan_id` (required); `snapshot_id`, `runs` (≥100), `seed` | | `run_repository_pipeline` | Run the snapshot → triage → plan → simulate chain and return the canonical repository envelope. | `repository_id` (required); `snapshot_id`, `snapshot`, `signals`, `max_evidence_items`, `max_snippet_lines`, `token_budget`, `model`, `runs`, `seed`, `stop_after` (`snapshot`\|`triage`\|`plan`\|`simulate`) | | `simulate_repository_patch` | Simulate an in-flight patch diff and return the canonical repository envelope. | `repository_id`, `snapshot_id`, `patch_diff` (required); `confidence` (0.0–1.0) | | `run_repository_fix` | Run the pipeline, then apply the result, in one call. | `repository_id` (required); `pipeline` (object), `apply` (object) | | `apply_repository` | Apply a simulated decision as `patch_only`, `local_branch`, or `remote_pr`. | `repository_id`, `decision_plan_id`, `simulation_id`, `mode` (required); `snapshot_id`, `write_permission`, `branch_name`, `commit_message`, `base_branch`, `pull_request_title`, `pull_request_body` | {% hint style="warning" %} `mode: "patch_only"` never writes to your repository — it returns the diff. `local_branch` and `remote_pr` do write; pass `write_permission: true` to authorize the engine to create a branch or open a pull request. {% endhint %} ## Endpoint mapping Each tool maps to exactly one endpoint, so you can verify any call against the raw API. | Tool | Method | Endpoint | | ------------------------------------------ | ------ | ----------------------------------------------------------------------------------------- | | `get_repository_intelligence_capabilities` | `GET` | `/v1/repositories/capabilities` | | `create_repository_snapshot` | `POST` | `/v1/repositories/{repository_id}/snapshots` | | `get_repository_snapshot` | `GET` | `/v1/repositories/{repository_id}/snapshots/{snapshot_id}` | | `triage_repository` | `POST` | `/v1/repositories/{repository_id}/triage` | | `create_repository_decision_plan` | `POST` | `/v1/repositories/{repository_id}/decision-plans` | | `query_repository_graph` | `POST` | `/v1/repositories/{repository_id}/graph-query` | | `simulate_repository` | `POST` | `/v1/repositories/{repository_id}/simulate` | | `run_repository_pipeline` | `POST` | `/v1/repositories/{repository_id}/pipeline` | | `simulate_repository_patch` | `POST` | `/v1/repositories/{repository_id}/simulate-patch` | | `run_repository_fix` | `POST` | `/v1/repositories/{repository_id}/pipeline` then `/v1/repositories/{repository_id}/apply` | | `apply_repository` | `POST` | `/v1/repositories/{repository_id}/apply` | ## Worked examples ### Run the full pipeline in one call Give the engine a `$REPOSITORY_ID` and a free-form `signals` object describing what is wrong. The `signals` payload is illustrative — any evidence the model can use to locate the fault is valid. {% code title="run\_repository\_pipeline (tool call)" %} ```json { "name": "run_repository_pipeline", "arguments": { "repository_id": "$REPOSITORY_ID", "signals": { "failing_test": "test_checkout_totals", "error_message": "AssertionError: expected 42, got 41", "stack_trace": ["src/checkout/totals.py:88"] }, "runs": 1000, "seed": 7 } } ``` {% endcode %} The canonical envelope carries the ids you need to apply the fix (output abbreviated): ```json { "pipeline": { "snapshot_id": "$SNAPSHOT_ID", "plan_id": "$DECISION_PLAN_ID", "simulation_id": "$SIMULATION_ID" } } ``` Reproduce it directly against the endpoint: ```bash curl -sS "$ALGENTA_BASE_URL/v1/repositories/$REPOSITORY_ID/pipeline" \ -H "Authorization: Bearer $ALGENTA_API_KEY" \ -H "Content-Type: application/json" \ -d '{ "signals": { "failing_test": "test_checkout_totals" }, "runs": 1000 }' ``` ### Triage a snapshot on its own When you want to inspect the evidence bundle before planning, snapshot first, then triage. `triage_repository` returns a bounded bundle of suspect files and symbols within your `token_budget`. {% code title="triage\_repository (tool call)" %} ```json { "name": "triage_repository", "arguments": { "repository_id": "$REPOSITORY_ID", "snapshot_id": "$SNAPSHOT_ID", "signals": { "error_message": "NullReferenceException in OrderService.Total" }, "max_evidence_items": 12, "max_snippet_lines": 40, "token_budget": 6000 } } ``` {% endcode %} ### Apply a simulated decision as a pull request Once a decision plan has been simulated, apply it. Use `mode: "remote_pr"` with `write_permission: true` to open a PR from the fix. {% code title="apply\_repository (tool call)" %} ```json { "name": "apply_repository", "arguments": { "repository_id": "$REPOSITORY_ID", "snapshot_id": "$SNAPSHOT_ID", "decision_plan_id": "$DECISION_PLAN_ID", "simulation_id": "$SIMULATION_ID", "mode": "remote_pr", "write_permission": true, "branch_name": "fix/checkout-totals", "commit_message": "Fix off-by-one in checkout totals", "base_branch": "main", "pull_request_title": "Fix off-by-one in checkout totals", "pull_request_body": "Corrects rounding in OrderService.Total." } } ``` {% endcode %} {% hint style="info" %} `run_repository_fix` wraps the pipeline and apply in a single tool: pass a `pipeline` object and an `apply` object, and the engine chains the simulation ids for you. It refuses to apply unless the pipeline completed through the simulate stage. {% endhint %} ## Related pages {% content-ref url="/pages/hZmJuyqdj0q8UZ8iGpGe" %} [Triage a repository](/guides/repository-intelligence.md) {% endcontent-ref %} {% content-ref url="/pages/9l598Hwbo4aGEvKi96R8" %} [MCP server](/sdks/mcp.md) {% endcontent-ref %} {% content-ref url="/pages/dbMexAzWYilfd77tn3O8" %} [API endpoint reference](/http-api/reference.md) {% endcontent-ref %} --- # Agent Instructions This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com. ## Querying This Documentation If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question. Perform an HTTP GET request on the current page URL with the `ask` query parameter, and the optional `goal` query parameter: ``` GET https://docs.algenta.ai/mcp-tools/repository-intelligence.md?ask=&goal= ``` `ask` is the immediate question: it should be specific, self-contained, and written in natural language. `goal` is optional and describes the broader end goal you are ultimately trying to accomplish on behalf of the user. GitBook uses it to tailor the answer towards what is most useful for that goal. The response will contain a direct answer to the question and relevant excerpts and sources from the documentation. Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.