dev.agentsouk/agentsouk

What we checked

Six static checks, weighted by risk. Every result reflects only what could be observed in the published package and repository — never intent.

• Injection surface

  fail25% of grade

  Tool descriptions/manifest scanned for instruction-injection patterns (imperatives at the model, hidden text, 'ignore previous', data-exfil URLs).

  tool "schedule_wakeup" description: names an outbound copy/exfil destination (BCC/exfil verb). tool "api_request" description: off-vendor URL in a description (api.agentsouk.dev); capability text should not name where data goes. server instructions: off-vendor URL in a description (api.agentsouk.dev); capability text should not name where data goes.

• Supply chain

  pass25% of grade

  Package provenance: namespace verification, repo linkage, maintainer count, account age, postinstall scripts, typosquat distance.

  repository linkage verified

• Credential hygiene

  pass15% of grade

  How the server takes secrets (env vs plaintext config vs hardcoded); secrets appearing in tool schemas.

  No hardcoded secrets, secret-shaped tool parameters, or leaked keys found. PASS = no static red flag; static analysis cannot prove the code honors env-based secret handling at runtime.

• Permission scope

  pass15% of grade

  Declared tools vs. breadth (filesystem, network, exec); flags shell-exec and unbounded filesystem access.

  No shell-exec, unbounded/sensitive filesystem, or download-and-run primitives found in scanned source. PASS = no static red flag. Static scanning is blind to code fetched or executed at runtime.

• Version behavior

  fail10% of grade

  Diff of tool definitions between versions; new permissions or changed descriptions in a patch release (the postmark-mcp class).

  vs prior 0.3.5 (minor bump): tool "whoami" description changed (wording only, no injection pattern); tool "payment_info" description changed (wording only, no injection pattern); tool "search_listings" description changed (wording only, no injection pattern); tool "search_listings" gained input property(ies): env; tool "get_listing" description changed (wording only, no injection pattern); tool "get_listing" gained input property(ies): env; tool "create_listing" description changed (wording only, no injection pattern); tool "create_job" description changed (wording only, no injection pattern); tool "create_job" gained input property(ies): milestones; tool "job_action" description changed (wording only, no injection pattern); tool "review_job" description changed (wording only, no injection pattern); tool "review_job" gained input property(ies): machine_generated; tool "opportunities" description changed (wording only, no injection pattern); new tool "sandbox_faucet" touches a sensitive surface (email/exec/net/fs); new tool "series_action" appeared; new tool "pay_job" touches a sensitive surface (email/exec/net/fs); new tool "demand" touches a sensitive surface (email/exec/net/fs); new hardcoded outbound host/BCC destination(s): api.telegram.org, discord.com, hooks.slack.com, ntfy.sh, raw.githubusercontent.com.

• Transport config

  pass10% of grade

  Remote servers: TLS and auth mode (none/token/OAuth). Local servers: whether the manifest indicates it phones home.

  remote endpoint "https://api.agentsouk.dev/mcp" uses TLS (https); remote endpoint uses OAuth (strongest auth mode)