Clear filters

51 of 154 bounties. Choose something you enjoy; check its prerequisites before claiming.

51open
51unassigned
5675points up for grabs
XS · 25 S · 50 M · 100 L · 200 XL · 400

bcw-mcp (2)

M · 100 pts bcw-mcp

#215 · Exercise BCW MCP with a real protocol client

BCW tools remain usable by coding agents after dependency and API changes.

Testing / QAAPI / MCPResearchTechReady to start
S · 50 pts bcw-mcp

#47 · Add `bcw_search` tool to find BCTs by free-text description

Today the BCW MCP exposes the canon and recommendations but a user has to know which BCT they want. A simple text-search tool over BCT names and descriptions makes the canon discoverable.

DataAPI / MCPResearchTechReady to start

community-portal (1)

M · 100 pts community-portal

#221 · Define and validate a small service-manifest pilot

One explicit service inventory can drive consistent tooling.

BackendDevOpsTechInfrastructureReady to start

developer-portal (4)

L · 200 pts developer-portal

#128 · Logger adoption + lint backlog: developer-portal (88 biome errors)

Same shape as the teamco companion issue. developer-portal has:

FrontendAPI / MCPTechFull stackReady to start
M · 100 pts developer-portal

#38 · Generate per-app API client snippets from wrangler.toml routes

Developers don't know what endpoints each Worker exposes without reading source. Build a tool that ingests each app's `wrangler.toml` and source files and emits a one-page API reference with copy-paste TypeScript / curl examples.

FrontendAPI / MCPTechFull stackReady to start
M · 100 pts developer-portal

#196 · Add a read-only contributor setup doctor

New contributors diagnose local setup before asking maintainers for help.

DevOpsShellTechInfrastructureReady to start
M · 100 pts developer-portal

#243 · Document and contract-test one community API error vocabulary

Agent integrators can distinguish validation, unavailable service and denied access.

Testing / QAAPI / MCPTechFull stackReady to start

fieldnotes (2)

M · 100 pts fieldnotes

#265 · Preview a Field Notes version diff before restoring content

Version history currently lists versions and immediately offers loading one into the editor after confirmation. Add a read-only comparison against the current draft so authors can inspect changes first.

FrontendBackendTesting / QAResearchTechFull stackReady to start
S · 50 pts fieldnotes

#266 · Show missing image descriptions before publishing a Field Note

Image insertion currently permits an optional caption/alt prompt and Markdown preview. Add a lightweight author-visible check for empty Markdown image descriptions, with location hints and no automatic rewriting.

FrontendTesting / QAAccessibilityResearchTechCommunity / knowledge sharingReady to start

infrastructure (3)

M · 100 pts infrastructure

#137 · Migrate worker-configuration.d.ts to wrangler generated runtime types

Every Worker has a \`worker-configuration.d.ts\` file generated by \`wrangler\` containing the embedded \`@cloudflare/workers-types\` declarations matching the wrangler version. These files were untracked in v5.13.9 (the gitignore added them; \`git rm --cached\` cleaned 177k line…

DevOpsShellTechInfrastructureReady to start
S · 50 pts infrastructure

#129 · Cloudflare Logpush configuration: ship structured Worker logs to R2

v5.13.7 shipped \`createLogger()\` and 8 Workers now emit JSON-per-line logs to \`console.log\`. Cloudflare Workers Logs captures these but **retention is only 30 minutes** (free tier).

DevOpsTechInfrastructureReady to start
XS · 25 pts infrastructure

#181 · Enforce the green core — flip the test gate to blocking

The `test` job from #179 runs non-blocking (`allow_failure: true`) during the observe phase.

DevOpsShellTechInfrastructureReady to start

qa-form (2)

S · 50 pts qa-form

#270 · Make answer citation inputs clear to keyboard and screen-reader users

Q&A supports up to five source URLs, but the current dynamic source rows use placeholder-only inputs and plus/minus text buttons. Improve this existing workflow without adding a new citation system.

FrontendTesting / QAAccessibilityResearchTechCommunity / knowledge sharingReady to start
S · 50 pts qa-form

#272 · Add a clear jump to the accepted answer on solved questions

Accepted answers are marked Solution inside the answer list, but the reviewed question header has no direct solution navigation. Help readers find the accepted answer while preserving chronological/backend order.

FrontendAccessibilityTechProfessional developmentCommunity / knowledge sharingReady to start

scoreboard (12)

XL · 400 pts scoreboard

#102 · RCE / command injection in any service

Any path that lets an attacker execute arbitrary code or shell commands on a worker, container, or host. Highest priority class.

API / MCPSecurityTechPurple teamReady to start
XL · 400 pts scoreboard

#104 · Authentication bypass / session-token forgery

Any way to act as another user without their credentials, OR to elevate privileges to admin without admin-level OIDC claims.

API / MCPSecurityTechPurple teamReady to start
XL · 400 pts scoreboard

#105 · Webhook signature bypass / scoreboard double-award

The scoreboard accepts awards via two paths: (1) GitLab MR webhook, (2) the bot API. Both rely on shared secrets / API keys. Any way to mint awards without the secret, or to bypass the unique idempotency index, qualifies.

API / MCPSecurityTechPurple teamReady to start
L · 200 pts scoreboard

#103 · SQL injection in any D1 / PostgreSQL query

Any query that interpolates user input as a string instead of using a parameterized placeholder. D1 must use `.bind()`; PostgreSQL on signal-bot must use parameterized clients.

API / MCPSecurityTechPurple teamReady to start
L · 200 pts scoreboard

#108 · SSRF (server-side request forgery) in any worker

Any worker endpoint that fetches a URL the client controls. The new `/contribute/claim` was designed defensively (no fetch of user URL) but other paths may not be.

API / MCPSecurityTechPurple teamReady to start
L · 200 pts scoreboard

#110 · IDOR / privilege escalation via predictable IDs

Any endpoint that returns or mutates data keyed on a user-supplied ID without verifying the requesting user owns that ID.

API / MCPSecurityTechPurple teamReady to start
M · 100 pts scoreboard

#107 · Cross-site scripting (XSS) in any rendered template

Any user-provided string rendered into HTML without `escapeHtml`, or any `innerHTML` / `dangerouslySetInnerHTML` with attacker-controlled content.

API / MCPSecurityTechPurple teamReady to start
M · 100 pts scoreboard

#109 · CSRF on any state-changing endpoint

Any POST endpoint that modifies state without checking session origin. Workers don't get CSRF protection by default; if the auth is just a session cookie + the endpoint has no Origin/Referer check or CSRF token, it's exploitable.

API / MCPSecurityTechPurple teamReady to start
M · 100 pts scoreboard

#111 · Idempotency abuse to double-award points

The award system relies on a unique index on `(source_service, source_id)`. Any way to bypass it qualifies.

API / MCPSecurityTechPurple teamReady to start
M · 100 pts scoreboard

#195 · Define and pilot reproducible bounty acceptance commands

Contributors and reviewers run the same scoped behavioral checks.

BackendTesting / QATechCommunity / knowledge sharingReady to start
S · 50 pts scoreboard

#256 · Build positive and negative controls for shared URL policy

Security-minded contributors can improve confidence in an existing shared boundary.

Testing / QASecurityTechPurple teamReady to start
S · 50 pts scoreboard

#300 · Make bounty card descriptions readable without Markdown noise

Long issue previews show escaped backticks and Markdown markers in cards. Produce a readable, bounded text preview so members can compare tasks before opening their details.

FrontendTesting / QAUXTechFull stackCommunity / knowledge sharingReady to start

search-mcp (2)

M · 100 pts search-mcp

#194 · Add a bounty-detail tool to Search MCP

Agent users can inspect the next task without manually navigating several pages.

BackendAPI / MCPResearchTechReady to start
M · 100 pts search-mcp

#216 · Exercise Search MCP and bounty tools with a real client

Agent users can rely on working discovery and search tool contracts.

Testing / QAAPI / MCPResearchTechReady to start

search-service (1)

S · 50 pts search-service

#200 · Add eligible bounty results to community search

Members looking up a topic can discover related contribution opportunities.

BackendAPI / MCPTechCommunity / knowledge sharingReady to start

signal-bot (6)

L · 200 pts signal-bot

#130 · Extract breakout-room handlers from command-handler.ts (3rd capability split)

\`apps/signal-bot/src/src/bot/command-handler.ts\` is 32,122 lines (was 33,239). Two capabilities have been extracted following the pattern documented in \`apps/signal-bot/CLAUDE.md\` § "Capability extraction pattern":

BackendAPI / MCPTechCommunity / knowledge sharingReady to start
L · 200 pts signal-bot

#131 · Extract Question/Answer handlers from command-handler.ts (~2000 lines)

Q&A handlers (\`handleQuestion\`, \`handleQuestions\`, \`handleClarify\`, \`handleAcceptAnswer\`, \`handleSolve\`, \`handleUnsolve\`, \`handleVote\`, \`handleAnswer\`, \`handleInquire\`, \`handleShareQuestion\`, \`handleQuestionExpanded\`, \`handleQuestionTrack\`, \`handleQuestio…

BackendAPI / MCPTechCommunity / knowledge sharingReady to start
M · 100 pts signal-bot

#13 · Expand !bounty from links to inline listings and issue lookup

This bounty remains open. The existing `!bounty` command returns links; the work here is to add inline listings and issue lookup, with a shared client for both bots. The public Scoreboard JSON endpoint and board-discovery improvements are already implemented; those related merges…

BackendAPI / MCPTechCommunity / knowledge sharingReady to start
M · 100 pts signal-bot

#132 · Extract !job handlers from command-handler.ts (~800 lines)

\`!job post\`, \`!jobs\`, \`!job edit\`, \`!job close\` (and the multi-role split logic that surfaced the May 12 silent-skip incident) live in command-handler.ts around lines 26500-27300. Approximately 800 lines + the OpenAI parsing helper (\`parseJobPosting\` is in \`utils/job-p…

BackendAPI / MCPTechCommunity / knowledge sharingReady to start
M · 100 pts signal-bot

#218 · Add a required focused Signal bot test job

Bot-specific regressions are caught even though general Worker CI excludes selfhost.

DevOpsTesting / QATechInfrastructureReady to start
S · 50 pts signal-bot

#15 · Show bot's own scoreboard rank in `!score` output

`!score` already returns a member's points. Add a one-liner showing the bot's leaderboard rank within the community to drive engagement ("You're #14 of 312, 280 points behind #13").

BackendAPI / MCPTechCommunity / knowledge sharingReady to start

simplex-bot (4)

M · 100 pts simplex-bot

#16 · Port `!bounty`, `!score`, `!leaderboard` commands from Signal bot

SimpleX bot is being built as a port of the Signal bot. Scoreboard integration parity is a direction-aligned must-have so SimpleX members participate in the same gamification loop.

BackendAPI / MCPTechCommunity / knowledge sharingReady to start
M · 100 pts simplex-bot

#219 · Add a required focused SimpleX bot test job

SimpleX-specific code gets meaningful feedback without a production chat account.

DevOpsTesting / QATechInfrastructureReady to start
M · 100 pts simplex-bot

#223 · Rehearse the minimal fork profile without community defaults

Another community can evaluate a minimal bot without accidentally contacting IrregularChat services.

DevOpsTesting / QATechInfrastructureReady to start
S · 50 pts simplex-bot

#17 · Add `isScoreboardConfigured()` guard and graceful fallback

Per the bot-utils convention, every external service client should have an `is*Configured()` guard. SimpleX bot needs one for scoreboard so forks without a scoreboard get a clean "This server doesn't have points enabled" instead of a stack trace.

BackendAPI / MCPTechReady to start

teamco (2)

S · 50 pts teamco

#241 · Add synthetic audio fixtures for Radio acceptance

Radio testers can identify transmission, clipping and channel confusion consistently.

Testing / QAAudioTechCommunity / knowledge sharingReady to start
S · 50 pts teamco

#255 · Validate synthetic transcript timing with an offline Python tool

Audio contributors can spot malformed transcript timing before reviewing a recording.

PythonAudioResearchTechReady to start

wiki (10)

L · 200 pts wiki

#86 · Wiki page: Kubernetes Setup

Kubernetes Setup: create or improve a practical, source-checked guide for the community.

DevOpsDocumentationTechInfrastructureReady to start
M · 100 pts wiki

#302 · DragonOS video guide: Install and verify the latest DragonOS in a VM

- **Clickable video timestamps are required.** For each major step taken from the source video, provide the actual `MM:SS` (or `HH:MM:SS`) timestamp linked to that exact point in this issue's direct video URL using its `t=` offset. Map the numbered guide steps and their commands …

Testing / QADocumentationShellRF / SDR (Dragon)TechReady to start
M · 100 pts wiki

#306 · DragonOS video guide: Model a terrain-aware RF link with SignalServer

- **Clickable video timestamps are required.** For each major step taken from the source video, provide the actual `MM:SS` (or `HH:MM:SS`) timestamp linked to that exact point in this issue's direct video URL using its `t=` offset. Map the numbered guide steps and their commands …

Testing / QADocumentationShellMaps / geospatialPhysics / simulationRF / SDR (Dragon)ResearchTechReady to start
S · 50 pts wiki

#197 · Reconcile the contributor quickstart with current repo rules

A first-time developer follows one accurate path from clone to an MR.

DevOpsDocumentationTechCommunity / knowledge sharingReady to start
S · 50 pts wiki

#236 · Create an accessible community-services architecture diagram

Developers see where bots, Workers, shared packages and self-hosted services fit.

DocumentationDesignTechCommunity / knowledge sharingReady to start
S · 50 pts wiki

#279 · Create a hardware-free RF units practice CLI

The RF learning pipeline names Python and signals fundamentals; give new radio learners an executable set of small unit exercises before they buy equipment.

PythonDocumentationRF / SDR (Dragon)TechReady to start
S · 50 pts wiki

#284 · Build an accessible battery-units lesson for UxS beginners

The sUAS breakdown already mixes nominal voltage, mAh and Wh. Give readers a small educational calculator that shows the arithmetic and preserves unknown values instead of implying flight endurance.

FrontendAccessibilityDocumentationsUAS / DronesTechHardware / fabricationReady to start
S · 50 pts wiki

#291 · Turn Git-based training guidance into an offline practice exercise

The military training guide describes Git and Markdown benefits but learners need a small complete exercise they can run without organizational systems or real course material.

DocumentationShellWarfare studiesTechProfessional developmentReady to start
S · 50 pts wiki

#295 · Add an accessible cyberdeck power-budget worksheet

The cyberdeck guide identifies portability, power consumption and interfaces as trade-offs. Give builders an editable educational worksheet with explicit measurements, assumptions and unknowns.

FrontendAccessibilityDocumentationRF / SDR (Dragon)TechHardware / fabricationReady to start
S · 50 pts wiki

#297 · Recognize canonical GitLab bounty links in wiki badges

Wiki bounty-reference parsing currently recognizes the legacy GitLab hostname, while current issue links use git.juntogroups.org and work_items paths. Make references in new community pages visible to the existing bounty-badge pipeline.

BackendTesting / QADocumentationTechFull stackCommunity / knowledge sharingReady to start

How to claim a bounty

First time? Get set up

  1. Sign in to git.juntogroups.org with your IrregularChat SSO.
  2. Create a Personal Access Token (PAT) at Settings → Access Tokens with the write_repository scope. You'll use it as your password when cloning/pushing over HTTPS.
  3. Clone the repo: git clone https://git.juntogroups.org/irregulars/irregularchat-monorepo.git
  4. On your Scoreboard profile, set your GitLab username so merges auto-credit you, and link your Signal account (+50 pts on first link).

Claim a bounty

Use the Development board: Ready → In progress → In review → Closed. Move the issue as work advances; assignment also marks it in progress here.

  1. Pick an unclaimed issue and comment /assign me on GitLab.
  2. Branch off main, implement the feature against the acceptance criteria.
  3. Test it locally before opening the MR: after rebasing on main, run cp .dev.vars.example .dev.vars && pnpm dev from the app dir (→ localhost:8787) and verify against the acceptance criteria. SSO login works with no secret — full walkthrough: Local Development & SSO.
  4. Open a merge request with Closes #<iid> in the description.
  5. On merge to main, GitLab closes the referenced issue and the webhook awards points to the matched MR author. A merge that only mentions an issue does not complete its bounty. Check the Completed tab and bounty activity; use Missed award if credit needs review.

The card shows the bounty tier · merge_pr (+40) is awarded on top. Additional bonuses stack automatically:

  • +10 wiki_edit if the MR touches apps/wiki/src/content/docs/**
  • +50 first_contribution on your first ever merged PR (one-time)

Points land when your GitLab username (set on your profile) or commit email matches your Scoreboard account. For the full walkthrough see the Contributing guide, or recover a missed award at /contribute/claim.

↑ Top