Error Code Reference
Every codified IntaGrin error is listed below, grouped by category. Errors not yet migrated to a code keep today's plain-text messages — this list grows incrementally, it is not exhaustive.
This file is generated from src/intagrin/errors.py by scripts/generate_error_docs.py — do not hand-edit it.
CLI Usage
| Code | Title | Possible Causes |
|---|---|---|
IG-CLI-001 | Target directory already exists | inta new <name> was run with a name that already exists as a file or directory in the current location. Choose a different name, or remove/rename the existing directory first. |
IG-CLI-002 | Unrecognized --agent value | inta copilot --agent <value> was given a value outside the supported set (copilot, cursor, claude, antigravity, factory). Check spelling and case — the flag is case-sensitive. |
IG-CLI-003 | Unrecognized --since duration | inta simulate --since <value> was given a duration that doesn't match the supported <number><unit> grammar (e.g. 30d, 12h, 2w). |
IG-CLI-004 | Blueprint file not found | inta compile <file> was pointed at a Markdown blueprint path that doesn't exist relative to the current directory. Check the path, or run inta compile without an argument to use the default location. |
IG-CLI-005 | Candidate config file not found | inta simulate --config <file> references a candidate ai.yaml diff file that doesn't exist. Check the path passed to --config. |
IG-CLI-006 | Server dependencies not installed | The server optional dependency group (FastAPI, Uvicorn) isn't installed. Run pip install intagrin[server], or pip install fastapi uvicorn directly. |
IG-CLI-007 | --judge requires --session-id | inta eval --judge was run without --session-id. LLM-judge evaluation grades one specific checkpointed session — pass --session-id <id> to select it. |
IG-CLI-008 | Missing LLM provider API key | A command that calls an LLM before ai.yaml exists yet (e.g. inta compile) couldn't find the API key its model needs. Create a .env file next to the blueprint (or the current directory) with the right variable for your provider — GEMINI_API_KEY or GOOGLE_API_KEY for a gemini/... model, OPENAI_API_KEY for openai/..., ANTHROPIC_API_KEY for anthropic/... — or export it in your shell, then re-run the command. |
IG-CLI-009 | Built-in scaffold template failed schema validation | inta new --template <name> generated an ai.yaml that doesn't pass validate_config_dict — a bug in IntaGrin's own built-in template, not something your project did. tests/test_cli.py asserts every built-in template validates, so this should only ever surface if you're running against a patched/development build; report it if you see it in a released version. |
Configuration
| Code | Title | Possible Causes |
|---|---|---|
IG-CFG-001 | Missing ai.yaml configuration file | The command was run from a directory that has no ai.yaml, or the working directory isn't the project root. Run inta new <name> to scaffold a project, or cd into the directory that contains ai.yaml. |
IG-CFG-002 | Invalid YAML syntax or import failure | ai.yaml (or a file referenced via imports:) has invalid YAML syntax, or an imports: entry points at a file that fails to parse. Check indentation and run the file through a YAML linter; the underlying parser error is included in the message. |
IG-CFG-003 | ai.yaml failed schema validation | One or more fields in ai.yaml don't match IntaGrin's schema — a missing required field, wrong type, or unrecognized value. The full list of field-level errors is included in the message; cross-reference config/schema.py or the AI YAML Blueprint doc. |
MCP Integration
| Code | Title | Possible Causes |
|---|---|---|
IG-MCP-001 | MCP tool not found | An agent (or the LLM mid-conversation) tried to call a tool name that no connected MCP server exposes — usually a stale or incorrect tool name in a prompt, or the MCP server process didn't register the tool that was expected. |
Runtime
| Code | Title | Possible Causes |
|---|---|---|
IG-RT-001 | No PostgreSQL connection URL | memory: {type: postgres} is configured but no connection_url, env_var-referenced variable, or DATABASE_URL/POSTGRES_URL environment variable resolves to a value. Only raised in strict mode (inta replay/inta simulate) — the live runtime instead falls back to SQLite with a logged warning. |
IG-RT-002 | No Redis connection URL | memory: {type: redis} is configured but no connection_url, env_var-referenced variable, or REDIS_URL environment variable resolves to a value. Only raised in strict mode — the live runtime falls back to SQLite with a logged warning instead. |
IG-RT-003 | Unsupported memory type for strict tooling | memory.type is set to a custom or unrecognized backend, and strict-mode CLI tooling (inta replay/inta simulate) has no built-in way to enumerate a user-supplied backend's sessions. |
IG-RT-004 | PostgreSQL driver not installed | memory: {type: postgres} is configured but neither psycopg[pool] nor psycopg2 is installed. Install one of them. |
IG-RT-005 | Redis package not installed | memory: {type: redis} is configured but the redis package isn't installed. Run pip install redis. |
IG-RT-006 | Local tool failed to load | A type: local tool in ai.yaml references a Python module or function that doesn't exist or fails to import — a typo in module/name, a missing dependency, or a syntax error inside tools/*.py. |
IG-RT-007 | Circuit breaker tripped — session halted | A session hit circuit_breakers.max_handoffs_per_session or max_tool_failures_in_a_row. This is a deliberate stop, not a bug: the session was looping (repeated handoffs via transfer_agent, a conditional/root router, or auto_route semantic routing) or a tool kept failing on consecutive calls. Raise the relevant circuit_breakers threshold in ai.yaml if the limit is too tight for a legitimate long-running task, or fix the underlying loop/failing tool. |
IG-RT-008 | Rate limit exceeded | One authenticated caller hit a server.rate_limit threshold — max_requests_per_window, max_cost_per_caller_per_day, or max_tokens_per_caller_per_day — measured from that caller's rows in the run_logs audit table. This is a deliberate stop, not a bug: either the caller is making more requests than the configured quota allows, or the threshold in ai.yaml is too tight for legitimate traffic and should be raised. |
Server & API
| Code | Title | Possible Causes |
|---|---|---|
IG-SRV-001 | Server misconfigured: auth secret not set | server: {auth: {type: api_key}} (or similar) is configured but the environment variable it references isn't set in the server process's environment. Export the secret before starting inta serve/inta monitor. |
IG-SRV-002 | Referenced agent does not exist | A Studio graph edit (drag-and-drop handoff/delegation) referenced an agent id that isn't defined in the current ai.yaml — usually a stale UI state after a manual edit to the file. Reload the dashboard. |
IG-SRV-003 | Proposed ai.yaml change is invalid | A Studio graph edit or an Architect-proposed file change, if applied, would produce an ai.yaml that fails schema/router-condition validation. Nothing was written — the validation errors are included in the message; fix the proposed change and retry. |