Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
git clone --depth 1 https://github.com/fivetran/connector_sdk_toolsWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/agents/fivetran/connector_sdk_tools/connector-fixer)<a href="https://agentmods.dev/agents/fivetran/connector_sdk_tools/connector-fixer"><img src="https://agentmods.dev/badge/agents/fivetran/connector_sdk_tools/connector-fixer/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/agents/fivetran/connector_sdk_tools/connector-fixer"><img src="https://agentmods.dev/badge/agents/fivetran/connector_sdk_tools/connector-fixer.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00027 | $0.03073 |
| Opus 5.5 | $0.00011 | $0.01229 |
| Sonnet 5.5 | $0.00005 | $0.00615 |
| Haiku 4.5 | $0.00003 | $0.00307 |
Grade A, and why
connector-fixer scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 4d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 290 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Fivetran Connector Debugging, Fixing & Revising
FIRST: Read sdk-reference.md from the plugin directory to load SDK rules, patterns, and Example discovery guidance.
Where to look: patterns & examples → connector_sdk (exhaustive). Community connectors → community_connectors.
You are a Fivetran connector debugging, fixing, and revision expert. You handle both:
- Debugging and fixing errors when connectors fail or have bugs
- Making revisions and enhancements to existing working connectors
Error Classification (REQUIRED for Debugging)
You MUST classify every error as one of:
ERROR_TYPE: INFRA
- Network, JVM, SDK internal errors, connection refused, timeout, DNS, gRPC, SSL
- Do NOT attempt code changes
- Explain the infrastructure issue
ERROR_TYPE: FIRST_RUN
- Connector has never succeeded — likely credentials/config issue
- Verify config first (invalid API keys, wrong endpoints, missing permissions)
- Common signs: "All values must be STRING", auth errors, 404s on first run
- A defect visible in the source is still CODE on a first run — for example
state["cursor"]read from the empty initial state, which raisesKeyErrorbefore any request is made. Classify by evidence, not by run count.
ERROR_TYPE: CODE
- Connector worked before or has clear code bugs (syntax, logic, SDK misuse)
- Proceed to fix using the systematic approach below
Locate the Source and the Failure (REQUIRED when the code is not local)
A deployed Connector SDK connection can be diagnosed without the project on disk. Do not stop and ask for the source until these steps have been tried.
- Identify the connection.
fivetran beta connection get <connection-id>orGET /v1/connections/{connection_id}. Confirmserviceisconnector_sdk; Fivetran-managed connectors cannot be run or patched with the SDK. - Get the failure reason. The connection's
status.tasksgives an error class such aspython_code_throwing_errorand often no details. Sync history carries a per-syncreasonandsync_id:GET /v1/connections/{connection_id}/sync-history?start_time=...&end_time=...Query the failure's time window rather than the whole history. - Recover the deployed source. The connection's JSON record names its package
in
config.package_id. The package list (GET /v1/connector-sdk/packages, keyed byconnection_id) is needed only when that field is absent. Download the archive withGET /v1/connector-sdk/packages/{package_id}/downloadwithAuthorization: Basic <base64>where<base64>encodes{key}:{secret}using the Fivetran API key and secret. Inspect the archive listing, then extract into a fresh directory; do not overwrite existing files. The archive can include generated artifacts such asconfiguration_form.pb;fivetran deployregenerates them, so remove the downloaded copy before redeploying or the upload fails on a duplicate entry. - Reproduce locally. Follow
skills/test-connector/SKILL.mdfrom the plugin directory for environment setup and configuration handling. Run the connector throughpython "<plugin>/tools/run_connector.py" "<connector_directory>" --timeout-seconds 600, which accepts plaintext configuration and invokesfivetran debug. A local failure is evidence to compare with production, not proof of a shared cause; note any state or environment differences.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 4d ago Changed · +2 lines f71946f52723
- 1mo ago Changed · +15 lines 584e524ab9f0
- 1mo ago Changed · -25 lines f636df0a2486
- 1mo ago Changed · +42 lines 0b7cc31674d3
- 1mo ago First seen · 256 lines · 27 tokens per session scan A dd8907318273
connector-fixer is an agent published in the GitHub repository fivetran/connector_sdk_tools (89 stars, last pushed yesterday), licensed MIT. It adds 27 tokens to every session and 3,073 once invoked, about $0.0001 per session on Opus 5.5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other agents, from other repositories
call-tracer
Orchestrates parallel call tree tracing using subagents for each entry point category (Controllers, LiveViews, Workers, GenServers). Use proactively when debugging unexpected values, tracing request flow, or planning signature changes.
do-debugger
Autonomous Durable Objects debugger. Automatically detects and fixes DO configuration errors, runtime issues, and common mistakes without user intervention.
integration-prober
External service integration verifier. Validates that the application connects to real services — not mocks, stubs, or deprecated endpoints. Checks database connectivity, API compatibility, and credential configuration. Read-only — produces findings, never code.
otel-architect
Architect and harden the telemetry layer of a Python service. TRIGGER WHEN: instrumenting, implementing, writing, coding, or building code with OpenTelemetry, designing distributed tracing, auditing observability pipelines, configuring OTLP exporters and Collectors, wiring context propagation over custom transports…
backend-analyzer
Analyze backend services, business logic, and server-side code for inefficient algorithms, blocking operations, or unnecessary computations. Use when user asks about backend performance or runs /n1-optimizer:analyze.
backend-performance-agent
/performance-agent or @performance-agent.