|
| 1 | +# Contributing to the IPMR Toolkit |
| 2 | + |
| 3 | +Thank you for considering a contribution. This toolkit is a working |
| 4 | +reference for fact-checkers, OSINT investigators, and civil society |
| 5 | +staff in Indonesia, Laos, Malaysia, the Philippines, Sri Lanka, and |
| 6 | +Thailand. Corrections, additions, and reports from practitioners on the |
| 7 | +ground are how it stays accurate. |
| 8 | + |
| 9 | +This guide is for outside contributors. If you are reading the toolkit |
| 10 | +to use it, start at the [home page](docs/index.md) instead. |
| 11 | + |
| 12 | +## Reporting an issue |
| 13 | + |
| 14 | +Open a [GitHub issue](https://github.com/denis989/ipmr-toolkit/issues/new) |
| 15 | +when you find: |
| 16 | + |
| 17 | +- A factual error in a tool card, decision tree, or country page. |
| 18 | +- A broken external link, retired tool, or vendor URL that has moved. |
| 19 | +- A safety override (S1–S10) that does not match the toolkit's |
| 20 | + documented framing for the case you encountered. |
| 21 | +- A platform behaviour the toolkit describes that has materially |
| 22 | + changed (API closure, takedown policy shift, new tipline endpoint). |
| 23 | + |
| 24 | +Include the page URL, the specific quote or claim you are flagging, |
| 25 | +and — where possible — a source or screenshot we can verify. |
| 26 | + |
| 27 | +## Suggesting a new tool |
| 28 | + |
| 29 | +The toolkit's selection criteria are documented in the |
| 30 | +[methodology section](docs/methodology/selection-criteria.md). |
| 31 | +Read those before opening a tool suggestion. Briefly: a tool needs to be |
| 32 | +operationally accessible to the region's fact-checkers, must have a |
| 33 | +documented limitation profile, and must fit one of the toolkit's six |
| 34 | +pillar–tier cells. |
| 35 | + |
| 36 | +Open an issue with the label `new-tool` and include: |
| 37 | + |
| 38 | +- Tool name and vendor. |
| 39 | +- The pillar–tier cell you think it belongs to (or "not sure"). |
| 40 | +- One paragraph on why a regional newsroom would actually reach for it |
| 41 | + (operational access, language coverage, cost, infrastructure). |
| 42 | +- Any independent accuracy or audit reference you are aware of — |
| 43 | + vendor self-reporting alone is not sufficient. |
| 44 | + |
| 45 | +The maintainers will review and either open a card draft, request more |
| 46 | +information, or document why the tool was considered but not included. |
| 47 | + |
| 48 | +## Proposing a correction |
| 49 | + |
| 50 | +For changes you are confident about, open a pull request: |
| 51 | + |
| 52 | +1. Fork the repository and create a branch (`fix/tool-card-typo` or |
| 53 | + `update/cofact-thailand-features`). |
| 54 | +2. Edit the relevant Markdown file under `docs/`. |
| 55 | +3. Keep the change tightly scoped. One correction or one new tool per |
| 56 | + pull request makes review faster. |
| 57 | +4. Match the surrounding voice. Tool cards stay practitioner-focused |
| 58 | + and avoid promotional language; decision trees stay operational and |
| 59 | + avoid restating workflow logic in prose. |
| 60 | +5. If your change touches a citation or a date-stamped figure (pricing, |
| 61 | + accuracy benchmark, accessed-on dates), update the citation block at |
| 62 | + the bottom of the page. |
| 63 | +6. Open the pull request against `main`. The maintainers will review, |
| 64 | + suggest edits, or merge. |
| 65 | + |
| 66 | +For larger structural changes — a new section, a new decision tree, a |
| 67 | +new country page — open an issue first to discuss scope before |
| 68 | +investing in the draft. |
| 69 | + |
| 70 | +## Style guidance |
| 71 | + |
| 72 | +The toolkit reads as written by a working editor, not a marketing |
| 73 | +team. A few conventions: |
| 74 | + |
| 75 | +- Tool cards follow the structure of `docs/tool-cards/_template.md`. |
| 76 | +- Limitations come before How-to-access on every tool card — the reader |
| 77 | + must be able to rule a tool out before investing in setup. |
| 78 | +- Detector tools never carry a binary "is AI" verdict on their own. |
| 79 | + Independent accuracy framing is mandatory; vendor self-reporting is |
| 80 | + always wrapped with the qualifier the methodology section documents. |
| 81 | +- Dates use Month YYYY format for pricing and accessed-on stamps. |
| 82 | +- External links live in a Sources block at the bottom of the page, |
| 83 | + not embedded mid-prose, except where the link is itself the |
| 84 | + evidence. |
| 85 | + |
| 86 | +## Code of conduct and licensing |
| 87 | + |
| 88 | +The toolkit content is published under |
| 89 | +[Creative Commons Attribution 4.0 International](LICENSE) |
| 90 | +(CC BY 4.0). Contributions you submit are accepted under the same |
| 91 | +licence — when you open a pull request, you agree that your edits may |
| 92 | +be redistributed under CC BY 4.0 with attribution. |
| 93 | + |
| 94 | +When reusing toolkit material outside the repository, attribute as: |
| 95 | + |
| 96 | +> "Triaging and Countering AI-Powered Disinformation and Influence |
| 97 | +> Operations" — an IPMR practitioner toolkit, Internews Europe, 2026. |
| 98 | +> CC BY 4.0. |
| 99 | +
|
| 100 | +## Substantive editorial questions |
| 101 | + |
| 102 | +For questions about the toolkit's editorial scope, regional coverage, |
| 103 | +or how to interpret a specific tool card or decision tree, write to: |
| 104 | + |
| 105 | +- **Vino Lucero** — Senior Project Officer-Asia — |
| 106 | + `jlucero@internews.org` |
| 107 | + |
| 108 | +## Maintainer |
| 109 | + |
| 110 | +The toolkit was produced by **Denis Yagodin** under contract with |
| 111 | +Internews Europe (IPMR project — Enhancing Indo-Pacific Media |
| 112 | +Resilience) with financial support from the European Union. |
| 113 | + |
| 114 | +Repository: |
| 115 | +[github.com/denis989/ipmr-toolkit](https://github.com/denis989/ipmr-toolkit) |
0 commit comments