Reference

Version 1, in beta. Every path starts with /api/cli/v1, every request carries a key as a bearer token, and every answer is { data, error }. The whole description is also OpenAPI 3.1.

Agent API

The server operations behind the hosted MCP tools. New integrations should use these. lockpick_wait_for_scan polls GET /agent/scans/{scanId}; lockpick_evaluate_scan reads that same scan and applies the assessment policy in the client. These two composed tools have no separate HTTP endpoint. Scripts can poll the scan status and evaluate its completion time, coverage and open finding counts against their policy.

GET /agent/me

The person, workspace, scopes and expiry of this credential, and what it can start.

• Permission: read
• MCP tool: lockpick_whoami

data

GET /agent/allowance

Free scans left this month, their reset dates and credits.

• Permission: read
• MCP tool: lockpick_get_allowance

data

GET /agent/monitor

Browse current security intelligence and explain matches for authorized sites.

• Permission: read
• MCP tool: lockpick_monitor

Query

data

Each item

GET /agent/sites

A page of sites with their latest scans.

• Permission: read
• MCP tool: lockpick_list_sites

Query

data

Each item

POST /agent/sites

Add a site, or return the one with that address.

• Permission: sites
• MCP tool: lockpick_add_site

Body

data

GET /agent/sites/{siteId}/status

One site's standing: its latest scans and what to do next.

• Permission: read
• MCP tool: lockpick_site_status

data

POST /agent/sites/{siteId}/dependencies

Send the package names and exact versions a site pins, never code, and read which advisories affect them.

• Permission: sites
• MCP tool: lockpick_send_dependencies

Body

data

POST /agent/sites/{siteId}/pulse

Start a Pulse on a site.

• Permission: pulse
• MCP tool: lockpick_start_pulse

Body

data

POST /agent/sites/{siteId}/deep-verifications

Issue a DNS or file ownership proof before a Deep scan.

• Permission: deep
• MCP tool: lockpick_start_deep_verification

Body

data

POST /agent/sites/{siteId}/deep-verifications/{verificationId}/verify

Check a published ownership proof. Each check uses one of its attempts.

• Permission: deep
• MCP tool: lockpick_check_deep_verification

data

POST /agent/sites/{siteId}/deep-operations

Prepare a Deep scan and get the link where the person approves it.

• Permission: deep
• MCP tool: lockpick_prepare_deep

Body

data

POST /agent/deep-operations/{operationId}/start

Start a Deep scan the person approved.

• Permission: deep
• MCP tool: lockpick_start_deep

data

POST /agent/free-site-switch

Remove a site so this month's free Pulse can move to another.

• Permission: sites and pulse
• MCP tool: lockpick_switch_free_site

Body

data

GET /agent/scans

A page of scans, newest first, with filters.

• Permission: read
• MCP tool: lockpick_list_scans

Query

data

Each item

GET /agent/scans/{scanId}

One scan with its status and headline.

• Permission: read
• MCP tool: lockpick_get_scan

data

GET /agent/scans/{scanId}/findings

A page of findings of a scan.

• Permission: read
• MCP tool: lockpick_list_findings
• Send x-lockpick-agent-features: finding-code to receive each finding's code (LP- and six characters, as the web report shows it). Without the header the answer has no code, the shape older clients expect.

Query

data

Each item

GET /agent/scans/{scanId}/findings/{findingId}

One finding with its redacted evidence.

• Permission: read
• MCP tool: lockpick_get_finding
• Send x-lockpick-agent-features: finding-code to receive each finding's code (LP- and six characters, as the web report shows it). Without the header the answer has no code, the shape older clients expect.

data

GET /agent/scans/{scanId}/repair-brief

One part of a scan's repair brief.

• Permission: read
• MCP tool: lockpick_get_repair_brief

Query

data

GET /agent/scans/{scanId}/comparison

What changed between a scan and the one it rechecks.

• Permission: read
• MCP tool: lockpick_compare_scans
• Send x-lockpick-agent-features: finding-code to receive each finding's code (LP- and six characters, as the web report shows it). Without the header the answer has no code, the shape older clients expect.

data

POST /agent/scans/{scanId}/dns-rechecks

Run the free DNS check again for a scan. Sends nothing to the site.

• Permission: pulse
• MCP tool: lockpick_recheck_dns

Body

data

GET /agent/scans/{scanId}/dns-rechecks/latest

A scan's latest DNS check again.

• Permission: read
• MCP tool: lockpick_get_dns_recheck

data

POST /agent/quick-checks

Start a free quick check of a homepage.

• Permission: pulse
• MCP tool: lockpick_quick_check

Body

data

POST /agent/repository-checks

Check a public GitHub repository (a skill, plugin, MCP server or app) before installing it.

• Permission: pulse
• MCP tool: lockpick_check_repository

Body

data

GET /agent/repository-checks/{checkId}

A repository check and its findings.

• Permission: read
• MCP tool: lockpick_get_repository_check

data

GET /agent/quick-checks/{checkId}

A quick check and its result.

• Permission: read
• MCP tool: lockpick_get_quick_check

data

GET /agent/docs/search

Search Lockpick's public docs.

• Permission: read
• MCP tool: lockpick_search_docs

Query

data

GET /agent/docs/page

One public docs page as Markdown, in parts when it is long.

• Permission: read
• MCP tool: lockpick_read_doc

Query

data

Device sign-in

How lockpick login gets a key: it asks for a code, the person approves it in the browser, and polling returns the key.

POST /device/code

Start a device sign-in and get a code for the person to approve in the browser.

• Permission: none, part of sign-in

Body

data

POST /device/token

Poll a device sign-in until the person approves it, then receive the key.

• Permission: none, part of sign-in

Body

Legacy

The original CLI routes, which CLI 0.2 uses. They stay as they are, and new integrations should use the Agent API above.

Original CLI routes

POST /projects/{projectId}/local-assessments/authorize

Authorize a local assessment before repository inspection. No source or results are uploaded.

• Permission: local

Body

Field	Type	Required
cliVersion	string	Yes
scannerVersion	string	Yes
rulesetVersion	string	Yes
reportSchemaVersion	string	Yes
checks	list of tracked-secrets, public-environment, browser-artifacts, supabase-client, supabase-migrations, git-history, docker-files, docker-values, actions-injection, source-maps	Yes

data

Field	Type	Always there
admissionId	string	Yes
projectId	string	Yes
workspaceId	string	Yes
policyVersion	string	Yes
startBy	string	Yes
checks	list of object	Yes

GET /me

The person, workspace and key behind this credential.

• Permission: read

data

Field	Type	Always there
user	object	Yes
workspace	object	Yes
key	object	Yes

DELETE /me

Revoke the presented key.

• Permission: any valid key

data

Field	Type	Always there
revoked	boolean	Yes

GET /projects

Sites (projects) with their latest Pulse and Deep scans.

• Permission: read

Each item in data

Field	Type	Always there
id	string	Yes
name	string	Yes
status	string	Yes
baseDomain	string or null	Yes
latestPulse	object or null	Yes
latestDeep	object or null	No
latestBox	object or null	Yes
webPath	string	Yes
webUrl	string	No

GET /projects/{projectId}/status

One site's latest results and what changed.

• Permission: read

data

Field	Type	Always there
project	object	Yes
latestPulse	object or null	Yes
latestDeep	object or null	No
latestBox	object or null	Yes
lastComparison	object or null	Yes
caveats	list of string	Yes

POST /projects/{projectId}/pulse

Start a Pulse on a site, or a recheck of its latest one.

• Permission: pulse
• Answers 201 when it works

Body

Field	Type	Required
idempotencyKey	string	Yes

data

Field	Type	Always there
id	string	Yes
kind	pulse, box (Deep, legacy name)	Yes
projectId	string	Yes
target	string	Yes
status	queued, running, cancelling, succeeded, incomplete, failed, cancelled	Yes
mode	autonomous, authenticated, active or null	Yes
createdAt	string or null	Yes
finishedAt	string or null	Yes
webPath	string	Yes
assessmentLevel	pulse, standard, deep or null	No
headline	object	No
webUrl	string	No
suggestedPollAfterSeconds	integer or null	No
baselineRunId	string or null	Yes
open	object	Yes
findingCount	integer	Yes
progress	object or null	Yes
coverage	object	Yes
caveats	list of string	Yes
outcome	object or null	No
exposureCheck	object or null	No

POST /projects/{projectId}/deep

Start a release's Deep scan with a CI key made for this site with Deep allowed.

• Permission: deep
• Answers 201 when it works

Body

Field	Type	Required
idempotencyKey	string	Yes

data

Field	Type	Always there
id	string	Yes
kind	pulse, box (Deep, legacy name)	Yes
projectId	string	Yes
target	string	Yes
status	queued, running, cancelling, succeeded, incomplete, failed, cancelled	Yes
mode	autonomous, authenticated, active or null	Yes
createdAt	string or null	Yes
finishedAt	string or null	Yes
webPath	string	Yes
assessmentLevel	pulse, standard, deep or null	No
headline	object	No
webUrl	string	No
suggestedPollAfterSeconds	integer or null	No
baselineRunId	string or null	Yes
open	object	Yes
findingCount	integer	Yes
progress	object or null	Yes
coverage	object	Yes
caveats	list of string	Yes
outcome	object or null	No
exposureCheck	object or null	No

GET /runs

Pulse and Deep scans, newest first.

• Permission: read

Query

Field	Type	Required
projectId	string	No
kind	pulse, box (Deep, legacy name)	No
assessmentLevel	pulse, standard, deep	No
status	queued, running, cancelling, succeeded, incomplete, failed, cancelled	No
limit	integer	No

Each item in data

Field	Type	Always there
id	string	Yes
kind	pulse, box (Deep, legacy name)	Yes
projectId	string	Yes
target	string	Yes
status	queued, running, cancelling, succeeded, incomplete, failed, cancelled	Yes
mode	autonomous, authenticated, active or null	Yes
createdAt	string or null	Yes
finishedAt	string or null	Yes
webPath	string	Yes
assessmentLevel	pulse, standard, deep or null	No
headline	object	No
webUrl	string	No
suggestedPollAfterSeconds	integer or null	No

GET /runs/{runId}

One run with its headline and counts.

• Permission: read

data

Field	Type	Always there
id	string	Yes
kind	pulse, box (Deep, legacy name)	Yes
projectId	string	Yes
target	string	Yes
status	queued, running, cancelling, succeeded, incomplete, failed, cancelled	Yes
mode	autonomous, authenticated, active or null	Yes
createdAt	string or null	Yes
finishedAt	string or null	Yes
webPath	string	Yes
assessmentLevel	pulse, standard, deep or null	No
headline	object	No
webUrl	string	No
suggestedPollAfterSeconds	integer or null	No
baselineRunId	string or null	Yes
open	object	Yes
findingCount	integer	Yes
progress	object or null	Yes
coverage	object	Yes
caveats	list of string	Yes
outcome	object or null	No
exposureCheck	object or null	No

GET /runs/{runId}/findings

Findings of a run.

• Permission: read

Query

Field	Type	Required
severity	critical, high, medium, low, info	No
state	attention, information, unknown, pass	No
all	true, false	No

Each item in data

Field	Type	Always there
id	string	Yes
code	string	No
runId	string	Yes
kind	pulse, box (Deep, legacy name)	Yes
severity	critical, high, medium, low, info	Yes
state	attention, information, unknown, pass	Yes
category	string	Yes
title	string	Yes
meaning	string	Yes
fix	string or null	Yes
confidence	string	Yes
evidence	list of string	Yes
sources	list of string	Yes
steps	list of string	Yes
webPath	string	Yes
webUrl	string	No
verdict	confirmed, needs_review, not_actionable or null	No
certainty	high, medium, low or null	No
exploitability	direct, conditional, hygiene, none or null	No
proof	object or null	No
proofGap	string or null	No
urgency	now, week, later, none	No
howSure	confirmed, likely, possible or null	No

GET /runs/{runId}/findings/{findingId}

One finding with its redacted evidence.

• Permission: read

data

Field	Type	Always there
id	string	Yes
code	string	No
runId	string	Yes
kind	pulse, box (Deep, legacy name)	Yes
severity	critical, high, medium, low, info	Yes
state	attention, information, unknown, pass	Yes
category	string	Yes
title	string	Yes
meaning	string	Yes
fix	string or null	Yes
confidence	string	Yes
evidence	list of string	Yes
sources	list of string	Yes
steps	list of string	Yes
webPath	string	Yes
webUrl	string	No
verdict	confirmed, needs_review, not_actionable or null	No
certainty	high, medium, low or null	No
exploitability	direct, conditional, hygiene, none or null	No
proof	object or null	No
proofGap	string or null	No
urgency	now, week, later, none	No
howSure	confirmed, likely, possible or null	No

GET /runs/{runId}/repair-brief

A run's fixes as one Markdown brief.

• Permission: read

Query

Field	Type	Required
findingId	string	No
agent	claude-code, codex, cursor, vscode, opencode, lovable, bolt, replit, v0	No

data

Field	Type	Always there
runId	string	Yes
findingIds	list of string	Yes
saved	boolean	Yes
markdown	string	Yes
webUrl	string	No

GET /runs/{runId}/comparison

What changed between a run and the one before it.

• Permission: read

data

Field	Type	Always there
runId	string	Yes
baselineRunId	string	Yes
error	string or null	Yes
changes	list of object	Yes
webUrl	string	No