Free: the Indicator Design & Tracking skill — the one that catches an invented indicator. Get it free →
Built for evaluators who sign their name to the report

Your AI doesn't know M&E.
These 39 skills fix that.

Install them into ChatGPT, Claude or Gemini. Your AI stops inventing indicators, stops guessing at donor formats, and says “that's not in the document” instead of making something up.

Or join the library — $97/year, with the first-report promise.

✓ One email. Instant download. No account.

✓ Built from FCDO, EU, UN, World Bank and ADB guidance.

✓ Nothing leaves your account.

Indicator Design & Tracking Skill loaded
You

What's FCDO's required indicator for adaptive programming?

Your AI

[NOT IN SOURCE]

You

Then what does the guidance say about adaptation?

Your AI

Adaptation is reported as a narrative under the VfM section, not as a standalone indicator. [EVIDENCE — §4.2]

No required metric is specified. [EVIDENCE — §4.2]

↳ A generic chatbot invents an indicator here. Yours refuses, then cites the section. That refusal is the product.
Formats kept current
FCDOEU / EuropeAidUN agenciesWorld BankADBOECD-DAC
You've already seen this fail

It doesn't fail loudly.
It fails confidently.

That is what makes it dangerous. A wrong answer looks exactly like a right one — until a donor reviewer asks where the number came from.

The four failure modes
01 / The invented indicator

“Percentage of beneficiaries with improved resilience.”

No definition. No baseline. No means of verification. It reads like an indicator and collapses under one question.

The tell: you can't name the document it came from.
02 / The guessed format

FCDO wants results. The EU wants intervention logic.

You get a generic logframe that satisfies neither, then spend a day reformatting it into two different templates.

The tell: it never asks which donor.
03 / The confident citation

A guideline it has never read, summarised in three bullets.

Every sentence sounds equally true. You cannot tell which came from the source and which came from the model.

The tell: no section numbers, no page references.
04 / The quiet omission

The finding that isn't in the draft.

Not an error — an absence. The inconvenient result, the caveat, the limitation. It disappears and nobody notices until the review.

The tell: the report is too clean.

The cost isn't the hour you saved.
It's the contract.

An evaluator's entire value is the sentence “you can trust this.” One unsupported claim in front of a donor doesn't cost you a correction. It costs you the next tender.

1 claimis all it takes
  • A challenged report means rework you don't bill for.
  • A questioned finding means a client who doesn't re-hire.
  • A pattern of both means you stop being shortlisted.
What a skill actually is

The folder you'd give a new M&E officer on day one.

Plain English, no code

Definitions, donor rules, worked examples, and the checks a reviewer runs. That's a skill.

Except your AI reads it in one second, cites the section it used, and refuses to answer when the source is silent.

A prompt gives you one answer.
A skill gives your AI the method.
  1. 01

    Download the folder

    One zip. The method, its reference files, a cheatsheet and a manifest that says where every rule came from.

  2. 02

    Drop it into your AI

    A Claude Project, a ChatGPT Custom GPT, a Gemini Gem — or your tool's skills folder. No code, no API, no setup.

  3. 03

    Ask in that domain

    It answers from the source and cites the section. If the source is silent, it says [NOT IN SOURCE] instead of inventing.

  4. 04

    Keep using it

    It stays in the folder. Every report, every logframe, every donor — the same method, without re-pasting anything.

The library

Six bundles.
Not 39 decisions.

Every skill belongs to a workflow you already run. Start with the one you're doing this week.

39 skills · 14 domains
01 · Design6 skills

A logframe the donor accepts

From a programme description to a defensible results chain with indicators, baselines and means of verification.

  • Theory of Change Development
  • Results Framework Development
  • Indicator Design & Tracking
  • GEDSI Indicators
  • Learning Agenda Development
  • Stakeholder Mapping
02 · Collect6 skills

Field data you can defend

Forms, sampling and quality checks that catch the failure before it reaches the dataset.

  • Kobo/ODK Form & QA
  • Sampling Methodology & Design
  • Survey & Questionnaire Design
  • Data Quality Assessment
  • Field Data Fraud Checks
  • Qualitative Data Methods
03 · Analyse5 skills

From 300 transcripts to a finding

Corpus setup, coding with an evidence trail, synthesis and independent verification.

  • Research & Evaluation Corpus Setup
  • Research & Evaluation Synthesis
  • Qualitative Coding (no software)
  • Data Verification & Validation
  • Value for Money Analysis
04 · Report7 skills

The report that survives review

Donor formats, report assembly, dashboards and the narrative each funder actually checks.

  • FCDO Reporting
  • EU Reporting
  • Donor Report Assembly
  • M&E Reporting & Documentation
  • Data Visualization & Dashboards
  • Power BI M&E Dashboard
  • Desk Review Generator
05 · Govern4 skills

AI that doesn't breach an NDA

Data protection, disclosure policy, inclusion and beneficiary feedback — the layer procurement asks about.

  • AI Governance & Data Protection
  • Gender & GESI Analysis
  • Beneficiary Feedback Systems
  • Adaptive Management & CLA
06 · Career4 skills

The next role

Interview scoring, donor CV conversion and portfolio artefacts that prove you can do the work.

  • M&E Interview Scorer
  • Donor CV Converter
  • Knowledge Management & Learning
  • Baseline & Endline Study Design
Also included · not the product

Twelve one-hour courses,
because sometimes you want it explained.

A skill loads the method. A course walks you through it once, on screen, ending in a file you keep. They're included because the best members read one and install the other.

Worth $348 separately
01

Build a Theory of Change & Logframe in an Hour

Blank page to a one-page ToC and a donor-ready logframe, with indicators, baselines and means of verification.

50 min · 143 slides
02

Indicators That Survive Review

From a weak activity count to an indicator matrix a donor accepts — definitions, disaggregation and MoV.

62 min · 132 slides
03

Donor Report Assembly in an Hour

The results table, the per-output narrative and the value-for-money note — in FCDO, EU, UN and World Bank formats.

65 min · 109 slides
04

Pass the M&E Interview in an Hour

Decode the ToR, build six STAR answers, survive the five take-home formats, rehearse with an AI panel.

61 min · 105 slides

Plus eight more, including Sampling & Data Quality, M&E Systems, AI Governance, Power BI Dashboards, Kobo/ODK, Qualitative Coding, Field Data Fraud Checks and Evaluation Plan. See all 12 courses →

Show, don't claim

What's actually in the folder.

No black box. Every skill ships with its method, its sources, and the rules that stop it inventing.

Evidence you can inspect
SKILL.md — excerptindicator-design
# Answering rules
1. Answer only from the files in this folder.
2. Cite the section for every claim.
3. Label authority vs reasoning.
4. Never invent a figure, rule or citation.
5. If the source is silent, reply [NOT IN SOURCE].
6. Ask before assuming a donor format.

# Required fields for every indicator
definition · unit · baseline · target ·
disaggregation · means of verification · frequency
Reviewer checklist — six questionsships with every skill
  • Can every number be traced to a named source, with a date?
  • Is every causal claim marked as contribution, not attribution?
  • Does each indicator have a unit, a baseline and a means of verification?
  • What did the source contain that the draft leaves out?
  • Which sentences are inference presented as evidence?
  • Would you sign this with your name and your client's name on it?
Who it's for

Pick the one that sounds
like your week.

Every page on this site is written for one job, in that person's words. These three are who the library was built for.

Start where you are
01 · Solo practice · 5–25 years

Independent evaluators & consultants

“My name is on the report and I can't afford an error with it.”
Billing $110–170/hr. Using AI in client work. The strongest M&E skill set and the least time.
Start with: Design + Report
Open the consultant page →
02 · The only M&E person

NGO M&E officers & managers

“The donor wants the logframe and the quarterly report — and I'm the only M&E person.”
Mostly Global South. Owns the logframe, the tracking and the donor report. The process is still manual.
Start with: Design + Report + Collect
Open the officer page →
03 · 10–200 staff

TPM & evaluation firms

“Twenty consultants, twenty AI habits, and a client asking who verified what.”
Delivering to FCDO, EU, UN and World Bank. Standardisation is the whole problem.
Start with: all six bundles + team
Open the firm page →

Also built for grants managers, field officers and jobseekers. See all six →

Buy it if…

  • You already use ChatGPT, Claude or Gemini for M&E work.
  • Your name is on the report and a donor or client will read it.
  • You have more than one donor format to satisfy from one dataset.
  • You can't paste beneficiary data into a public tool.
  • You want the method, not another AI course.

Skip it if…

  • You want the AI to write the report unreviewed.
  • You need a platform that stores and tracks programme data — that's TolaData or DevResults.
  • You're looking for a certificate for your profile.
  • You've never opened a chatbot and don't intend to.
  • You need a guarantee that a donor will approve. Nobody can give you that.
The offer

One library. One price.
Everything else is included.

No tiers to compare, no bundles to decode, no upsell after checkout. One membership, and the whole library is yours.

Founding · first 100

Everything you get

  • ✓All 39 skills across 14 M&E domains — every bundle, no exclusions.
  • ✓Every new skill, monthly — built from what members ask for, not what's easy to write.
  • ✓Quarterly model re-tests — we run the golden sets on each major release so your library doesn't quietly rot.
  • ✓Donor formats kept current — FCDO, EU, UN, World Bank and ADB, updated as they change.
  • ✓All 12 courses — the hour of instruction when you want it, included.
  • ✓Every template pack — logframe, indicator tracker, donor report, grant pack and the rest.
  • ✓Works in your AI — ChatGPT, Claude, Gemini, Copilot or a local model.

Three promises. No fine print.

Promise 01 · effort≈15 min to us

The 10-Minute Install Promise

If you can't get a skill running inside ten minutes, send us a screenshot and we'll install it with you on a short call. We'd rather spend fifteen minutes with you than have you own something you never opened.

Promise 02 · outcome≈1–2 hrs to us

The First-Report Promise

Run the library on one real report. If it doesn't catch at least one claim, number or omission you would have missed, send us the report and we'll run the verification and send you the ledger. You keep the library either way.

Promise 03 · exit$0 to us

The Keep-Everything Promise

Cancel whenever you like and keep every file you have downloaded. No clawback, no hostage-taking, no “access ends immediately.” You paid for the files; the files are yours.

And the honest part

We don't do no-questions-asked refunds on digital files. Once you've downloaded it, it's yours forever and we can't un-send it. So instead of a refund, you get the three promises above — each of which costs us more than a refund would and gets you further.

If none of them work for you, email us and we'll sort it out. We're not going to make money from someone who got nothing.

The M&E Skill Library Founding
first 100
39 installable skills14 domains, every bundle
$1,131
12 narrated coursesOne hour each, artefact included
$348
11 template packsEditable, with worked examples
$319
A new skill every month12 a year, built from member requests
$348
Quarterly model re-testsSo the library stays current
included
If bought separately$2,146
↳ Why only 100. Every founding install is handled one-to-one by our team, and we can only take 100. After that the price goes to $149 and installs move to the wider team.
$149 / year after founding
$97 / year
That's $8.08 a month. Cancel anytime.
Start my library — $97
One skill on its own is $29. Three skills is $87. The whole library is $97 a year.
Instant access · Secure checkout by Stripe
What happens after you join

A guarantee only counts
if you actually use it.

So this is the sequence. We're with you for all of it, and the promises are live from day one.

Your first 14 days
Day 0
Install

One screenshot, one instruction. Not working in ten minutes? Reply and we'll do it with you.

→ skill loaded
Day 1
First refusal

Paste one prompt about live work. Watch it say [NOT IN SOURCE] and cite the section instead.

→ you see it work
Day 3
Your real report

Run it on something due. Reply with the report type and we'll send a prompt tuned to it.

→ used on real work
Day 7
Your statement

Generate the verification statement from your own work. That's the page you hand the donor.

→ artefact in hand
Day 14
The check

Did it catch something? If not, send us the report. We'll run the verification.

→ promise honoured
The honest comparison

Why not just use the prompts you already have?

Prompt vs skill
A prompt pack A skill (this) A generic chatbot Hiring a consultant
Gives youOne answerThe whole methodWhatever it remembersThe answer, and the invoice
When the model updatesSilently breaksRe-tested quarterlyBehaviour changesUnaffected
Can you tell what it invented?NoCites the section or refusesNoYou hope so
Knows your donor formatOnly if you paste itBuilt in, five formatsGenericUsually
Time to first resultMinutesMinutesMinutesDays to weeks
CostFree$97 / yearFree–$20/mo$500–$2,000 per deliverable
Your dataStays with youStays in your accountDepends on settingsLeaves your organisation

You don't need to throw away the prompts. Keep them. The skills are what you use when the output has to survive a reviewer.

Free · no account

Start with the one that catches the invented indicator.

Indicator Design & Tracking builds SMART indicators, Indicator Definition Sheets and progress tracking — and refuses to invent a baseline when the source doesn't have one. One email. Instant download.

One free skill per person. The other 38 are in the library.

Instant download. Plus the free briefing. Unsubscribe anytime. Privacy.

Before you install

The honest answers.

Is this just a pack of prompts?

No. A prompt gives you one answer, and it silently stops working when the model changes. A skill carries the method, the reference files and the answering rules together. It can cite the section it used, and it can refuse when the source is silent.

Skills are also re-tested against each major model release, so the library doesn't quietly rot the way a downloaded prompt pack does.

What if it doesn't work for me?

You get three promises. If you can't install it in ten minutes, we'll do it with you on a call. If it doesn't catch something you'd have missed on your first real report, we'll run the verification. If you want out, cancel and keep every file.

We don't do no-questions-asked refunds on digital files, because once it's downloaded we can't un-send it. But if none of the promises work, email us and we'll sort it out.

Do I need to be technical?

No. If you can create a Claude Project, a ChatGPT Custom GPT or a Gemini Gem, you can install a skill. That is the whole procedure: download the folder, attach it, ask a question.

There's no code, no API key and no setup. The one-page install guide ships with every skill, and the free ones include a short video.

Does my client data leave my account?

No. Skills run inside your own ChatGPT, Claude or Gemini account. Evaluation AI never receives your documents, and there is no server in the middle.

The Governance bundle covers redaction, disclosure policy and the data-flow map your procurement team will ask for.

Which donor formats are covered?

FCDO, EU / EuropeAid, UN agency frameworks, World Bank and ADB — plus OECD-DAC criteria. The method skills (indicators, theory of change, sampling, qualitative coding) are donor-agnostic and work with any template.

If your donor uses a custom format, you adapt it once and the skill reuses it.

How is this different from TolaData, DevResults or ActivityInfo?

Those are platforms that hold your programme data and reports. We don't store anything. We are the method layer inside the AI you already use — the part that checks what the AI wrote before it reaches a donor.

Most members use both: a platform for the data, a skill for the thinking and the writing.

Do I have to subscribe, or can I buy one skill?

You can buy any single skill on its own for $29. The library is better value the moment you want two: three skills cost $87 and the whole library is $97 a year.

There's no contract. Cancel from your account and keep every file you've downloaded.

Two ways to start

Neither one can lose.

Take the free skill and test it on something real. Or take the library and let the first-report promise carry the risk. Both start the same way: you see it work before you decide anything.

Option A · free
One skill, no card

Indicator Design & Tracking. Install it, run it on one report, see it refuse to invent.

Send me the free skill
Option B · $97/year
The whole library

39 skills, 12 courses, every pack, a new skill monthly. First-report promise included.

Start my library — $97

Founding price for the first 100. Cancel anytime and keep every file.

Indicator Design — freeInstant download, no account Get it free