MCP reference · JoesProxy

Why is a response body missing?

Connect an enrolled client. Use documentation_search to find “missing body”, then documentation_read to read the inspector guidance. Check one captured request before attributing the omission to the upstream server.

Look up the name your agent sees. A tool can inspect or change something on the Mac; a resource returns information; a prompt gives the agent instructions. If a tool appears in discovery but the call is denied, check the capability on its reference page against this client’s grant.

Alphabetical index by exact name

Compute on the capture Mac

capture_calculate tool: Check the completion status, selected source and exact numeric summary. For prices 0.1 and 0.2 the expected sum is the decimal string "0.3".

Run a controlled media test

apply_hls_fixture_start tool: After local review, inspect the resulting run ID and fixture status before using the stream.
hls_fixture_stop tool: Stop the owned fixture run and unregister its local fixture origin.
hls_fixture_window_preview tool: Inspect the publication window at a particular elapsed time without fetching a segment.
hls_substitution_profiles_list tool: Read available test-stream profiles before choosing a substitution.
hls_substitution_status tool: Inspect owned substitutions and reported hits. A hit is not proof the player recovered.
hls_substitution_stop tool: End an owned stream substitution so matching requests can use the upstream source again.
propose_hls_fixture_start tool: Prepare a baseline test stream and inspect the returned proposal. Preparing is not serving.

Test network conditions

apply_change_throttle tool: Apply the reviewed proposal and repeat the same app action, then compare transfer timing.
propose_change_throttle tool: Prepare a paced-network comparison for your test device. Inspect the proposed preset before applying it.

Find help

documentation_list tool: Browse topic IDs before requesting a section. A page of results is not necessarily the whole catalog.
documentation_read tool: Read the inspector’s explanation of missing bodies and override attribution.
documentation_search tool: Find help topics about missing response bodies. Select a topic identifier from the results.

Check the workspace

agent_toolset_status tool: Inspect the connected client’s discovery selection. A selected toolset does not authorize changing traffic.
approvals_list tool: Inspect approvals visible to this client. Listing a request does not grant it.
capture_status tool: Inspect current capture limits and admission pressure without reading any response bodies.
client_cleanup_status tool: Review client-owned effects and whether the session can finish cleanly. This check does not stop anything.
list_discovered_devices tool: List devices seen by this capture. Correlate the intended device with a fresh app action.
runtime_status tool: Inspect service readiness and grants before choosing the next test.
setup_readiness tool: Read the target-specific checks and the next human action. Remote certificate trust remains a device-side check.

Inspect a capture

bookmark_get tool: Read one saved bookmark. Its historical response is not the current origin response.
bookmarks_list tool: Find a bounded page of saved bookmarks. Bookmark identity is different from capture session identity.
effect_bundle_list tool: List effect bundles retained for this client. A retained bundle does not mean its effects are active.
override_get tool: Inspect a single rule to identify its intended match. Check a new request for an actual hit.
overrides_list tool: Inspect the saved rule list before diagnosing an unexpected response.
session_summary tool: Inspect one captured request rather than exporting the whole session list.
sessions_list tool: List a bounded page of captured requests. Use an actual returned session identifier in subsequent reads.
sessions_search tool: Find a short page of matching GET requests without requesting response bodies.

Investigate media

hls_fixture_capabilities tool: Discover fixture routing and availability before picking a test stream.
hls_fixture_profiles_list tool: List the available synthetic profiles and select a baseline to compare against a fault.
hls_fixture_status tool: List owned fixture runs. With a specific run ID, check that experiment rather than all runs.
media_manifest_structure tool: Inspect the playlist or manifest structure captured in a request.
media_readiness tool: See which media investigations are available for this captured response.
media_related_sessions tool: Find related media requests in the capture. Their presence is network evidence, not proof of playback.
media_summary tool: Read the available media metadata for a selected session.
operation_status tool: Check progress of a previously started media operation. Request the full result only once it is ready.

Read workspace resources

joesproxy://agent/capabilities resource: Discover capabilities reported to this client. Discovery is not authorization.
joesproxy://agent/guide resource: Read the running server’s guidance for the client.
joesproxy://audit/recent resource: Review recent audit evidence visible to your client.
joesproxy://capture/health resource: Check capture health before assuming a client-side fault.
joesproxy://capture/summary resource: Read a capture overview without copying individual payloads.
joesproxy://docs/index resource: Locate documentation topics and catalog information.
joesproxy://server/status resource: Read server state before interpreting missing traffic.
joesproxy://setup/guide resource: Read setup guidance, then verify the actual target device separately.

Use a guided prompt

joesproxy_diagnose prompt: Get the diagnostic playbook. Treat its suggested investigations as guidance, not automatic approval.
joesproxy_integrations prompt: Read the companion integration guidance. Prompt text is not evidence that a companion is installed.
joesproxy_setup_onboarding prompt: Get the onboarding sequence for listener status, target readiness and device-side verification.